# Welcome to FLORA Docs Welcome to FLORA's documentation page. {% hint style="info" %} **Looking for support?** Email , or visit [Help & Support](/getting-started/help-and-support) to report a bug or request a feature. {% endhint %} ### FLORA Intro Here's a quick intro to FLORA and how it works. {% embed url="" %} ### FLORA Training Sessions We host weekly live sessions for FLORA users and subscribers. We walk through FLORA essentials and some advanced tips & tricks (and often share upcoming features!). We'd love to see you in an upcoming session and hear your feedback and questions. RSVP to FLORA Training Sessions ### Jump right in
Getting StartedSetting up for FLORA/files/MgMfPO8H4TBHwI075oru/pages/CyH2xJQs9yWJ1S8BYNav
NodesLearn the basics of Nodes/files/EpdECI1pNru6KrOttVSd/pages/cEHgDv7NGBquDGHeLyQ4
EditorExplore our Editor/files/6t9huKHUkiluZzmlJpkq/pages/JjjojIyKxaiBPzzLwvtg
CommunityExplore FLORA's creative community/files/tRmbArphQ2c2teqb1uBH/pages/CwUHvR1kix6rMS3t4phU
# Help & Support Need help with FLORA? You're in the right place. ## Contact Support For account issues, billing questions, technical problems, or anything else — email . ## Report a Bug Found something broken? Tell us what happened so we can fix it. Report a Bug ## Request a Feature Have an idea for FLORA? We'd love to hear it. Request a Feature ## Other Resources * [Quickstart](/getting-started/quickstart) — new to FLORA? Start here * [Common Use Cases / FAQs](/faqs-how-to.../common-use-cases-faqs) * [FLORA Training Sessions](https://luma.com/floraai?k=c) — weekly live walkthroughs * [Community](/more/community) — Discord and creative workflows # Quickstart FLORA philosophy, creative process, and how to get started! FLORA is the first AI-powered creative workflow system built for professionals who think in systems, not just outputs. Unlike traditional AI tools focused on one-off generations, FLORA empowers creative teams to design structured, scalable workflows that enhance speed, control, and real-time collaboration. This quickstart guide will show you how to harness the power of FLORA's node based system to create structured, scalable workflows that give you speed, control, and creative freedom. ## Sign Up & Set Up Sign in with Google or use email to create your FLORA account (you may need to verify your email address).
## Welcome to your creative canvas! Your customized workspace is now set up, ready to bring your ideas to life. Dive in, explore the editor, nodes, and start building your creative workflow. Happy creating! 🥚 Want to learn some FLORA core concepts before you start creating? Dive right into learning about our modality Nodes and Editor. [Node Overview](/nodes/editor) # FLORA Concepts An introduction to FLORA and our core concepts: ### An Infinite Canvas for your Creative Process FLORA isn’t just another AI tool—it’s a complete system designed to align with the natural flow of creativity. It allows professionals to ideate, iterate, and refine their creative work at the speed of their thoughts on an infinite canvas. ### Workflows, Not just Outputs Unlike conventional AI tools that focus solely on generating assets, FLORA enables you to build workflows. By connecting Nodes, you can shape, refine, and scale your creative process, turning them into creative operative systems. ### From Concept to Execution With FLORA, you can ideate, iterate, and refine. * **Ideate**: Brainstorm and conceptualize ideas, exploring different concepts and possibilities. FLORA provides a space for an organic creative process, allowing your thoughts to flow freely and take shape as they visualize. * **Iterate**: Experiment boldly, refining ideas through continuous exploration and evolution. FLORA fosters a dynamic environment where you can build upon initial concepts, integrate feedback, and push creative boundaries to dream. * **Refine**: Elevate your work to its highest fidelity by honing details and perfecting execution. FLORA empowers you with the tools to sharpen your vision, ensuring your final creation is polished, impactful, and presentation-ready. \ Read more about our philosophy in our FLORA [Manifesto](/getting-started/publish-your-docs/manifesto). ###
# Manifesto **Current AI creative tools are made by non-creatives for other non-creatives to feel creative.** We’re a team of creatives who founded FLORA to solve our own problem: the lack of creative control in AI. We believe AI creative tools should be more than toys for generating AI slop. We are obsessed with building a power tool that will profoundly shape the future of creative work. Traditional creative tools give you control, but are unintuitive & time-consuming. They take years to master, and we waste even more time doing rote, repetitive workflows. AI tools make it easy to create, but lack creative control. This is great for making AI slop, but not for doing great creative work. We are building FLORA so that creative professionals can have the best of both worlds: creative ease & creative control. We believe in maximizing creative agency by building a creative tool that’s both intuitive & powerful. We want to help creatives achieve flow state. Founded out of an art & technology graduate program called NYU ITP, FLORA is an applied HCI company with a mission to make it possible to speak your ideas into existence, with a high degree of creative control. We are called FLORA because creation should feel natural. Plant your ideas. Watch them grow. **Welcome to FLORA.** 🌸
# Our Product Philosophy High level conceptualization of FLORA as a creative tool. We think creating should feel natural. The creative process & nature are both inherently non-linear and organic. \ By treating each modality as an independent node capable of non-linear network, we liberate the creative process from the constraints of hierarchical perception. \ This approach allows for fluidity in arranging and rearranging the connections—or “noodles”—between nodes, enabling diverse visual content to emerge across mediums and temporalities. \ This dynamic structure supports divergent thinking for conceptual exploration and convergent organization for streamlined production and distribution. With hope for new fissures and syntheses, we spawned FLORA—not just a creative software but a creative evolution. # Node Overview In FLORA, Nodes are the core creative units — connect them to build your workflows. FLORA operates on a node-based system where users can interconnect different AI models for greater creative control. Read more about our approach in [Our Product Philosophy](/getting-started/publish-your-docs/our-product-philosophy). In the next few sections, we'll explore the core building blocks for your work in FLORA — the Text Node, Image Node, Video Node, and Audio Node — and see how each one can be utilized to achieve your desired output. ### Here's an overview of our Nodes: {% stepper %} {% step %} [**Text Node**](/nodes/text-node) {% endstep %} {% step %} [**Image Node**](/nodes/image-node) {% endstep %} {% step %} [**Video Node**](/nodes/video-node) {% endstep %} {% step %} ### [Audio Node](/nodes/audio-node) {% endstep %} {% endstepper %} ## Node Toolbar A toolbar appears upon hovering on nodes of all modalities, above each node. The toolbar is contextual and contents vary depending on the node type.
# Text Node Learn the basics of text nodes Here is a quick introduction to our Text Node from our founding designer Ethan. {% embed url="" %} Text Node is a powerful tool for concepting, visual analysis, and information processing in creation of your personalized workflows. Treat the text node as an information hub, get creative.
Explore the following sections: [Text to Text](/nodes/text-node/text-to-text), [Image to Text](/nodes/text-node/image-to-text), and [Video to Text](/nodes/text-node/video-to-text) to see how you could get creative with text node! *** ### Models Visit our [Text Models](/models/text-models) section to learn about the text models and capabilities available in the Text Node. # Text to Text ## Summary Text to text allows for textual visualization, information integration, and structured workflow organization.
## Prompt * Sample Prompts: * "An excerpt from a dystopian novel set in the future" * "A creative script for an experimental film scene" * "A surreal underwater scene with floating lanterns illuminating the deep" * "An eerie perspective from inside a music box as it slowly winds down" * "A surreal landscape where the sky melts into the ocean" * "Combine these two ideas"

Connecting two text nodes to one for information integration

## How to use Text to Text is a powerful tool in your creation process. Using Text to Text allows you to combine and transform infinite ideas quickly and efficiently. By taking one or as many text nodes as you like, connecting them to other text nodes, you can conceptualize, iterate, and synthesize. {% hint style="info" %} Try connecting multiple text nodes to one to synthesize ideas, or branching one node out to many to expand and iterate on your vision. {% endhint %} Here are some sample workflows that highlight different techniques using Text nodes to Text nodes: * [Copy Generator Workflow](https://www.florafauna.ai/view/d89c012b-9134-4d52-9ed8-86f4707b4b44) * [Marketing Ideas Workflow](https://www.florafauna.ai/view/dc6c1ad6-6233-4e61-910c-3288b6552c1e) * [Sketches to storyboard Workflow](https://www.florafauna.ai/view/e0161835-2c9c-4546-9781-bf1cc5dccf43) ## Prompt Splitter If you have a text node with an output that contains a list, your toolbar will automatically display an option to "Split list into nodes". This will break out the text node into multiple nodes, each containing one of the list items. This is perfect when you're using a text node to ideate creative directions or prompt ideas, and you want to break those ideas out into separate nodes to use as inputs for downstream image or video nodes. # Image to Text ## Summary By connecting an image node to a text node, you can extract information from visual content, style, and composition. You can also get creative in your ask to abstract the visual information from the image node.

"Give me a visual analysis on the visual style and composition of this image"

"Capture the mood of this image in a haiku"

## Prompt * "Describe this image in a few sentences", * "What hidden rhythm flows through this scene?", * "Give me a visual analysis of the composition of this image", * "Conjure a visual poem for this". *** # Video to Text ## Summary Video to text can be performed by connecting a Video Node with a Text Node. Similar to Image to text, it can be used to extract information from any video inputs. This combination of nodes can be used to take actions like **describe the video**, or **give a numbered list of frames with detailed description of visual content.**
## Prompt * **"**Describe this video in a few sentence**"** * "Describe the composition and focal points of each frame, on how elements are arranged and how they guide the viewer's attention." * "Illustrate the atmosphere of each scene, focusing on sensory details such as lighting, visceral color sensation, and spatial depth." * "Describe how recurring motifs and visual patterns contribute to thematic development in the video." * "Give me a list of frame in this video and describe each visual composition." # Audio to Text By connecting an audio node to a text node, you can transcribe speech, capture dialogue, and turn spoken content into text you can feed into the rest of your workflow. Once the audio is transcribed, you can ask a new text node to summarize it, pull out quotes, translate it, or extract any information worth keeping.
# Image Node Learn the basics of image nodes The Image Node enables visual elaboration, concept exploration, information layering, and structured visual composition. It serves as a bridge between textual and visual information, allowing you to iterate on design ideas, experiment with aesthetic styles, color palettes, and create complex visual narratives by combining multiple images into one.
## Models Visit our [Image Models](/models/image-models)section to learn about the image models and capabilities available in the Image Node. *** # Text to Image ## Summary Text to Image is a powerful tool for idea visualization. It empowers creators to visualize abstract ideas, experiment with styles, and communicate complex narratives visually. Whether you’re concepting a new design, prototyping marketing materials, or crafting immersive worlds, Text to Image provides limitless possibilities for creative exploration and visual storytelling. The Text to Image node generates an image based on text context. It uses Flux Dev as the default model. {% hint style="info" %} [Read more](/models/image-models) on the Text to Image models we offer. {% endhint %} You can interact with it in a number of ways: * Type in a text prompt and hit generate to get an image * Connect a text output into the node to populate the text prompt * Use the prompt improver button to either generate a prompt from scratch (empty prompt field) or improve your current prompt (populated prompt field).
# Image to Image ## Summary Image to Image is an essential tool for creative evolution. It empowers transforming existing visuals, refining details, experimenting with styles, and enhancing narratives with precision. Whether you’re iterating on a concept sketch, reimagining a scene with new aesthetics, or blending multiple ideas into a cohesive composition, **Image to Image** unlocks new dimensions of creativity, enabling continuous exploration and storytelling through visual transformation.
### Image to Image Models Visit [Image Models](/models/image-models) to see a full list of image to image (i2i) models and their capabilities. ### Parameters Image models generally accept the following parameters. Check each model for specifics. | Parameter | Type | Effect on Output | | ---------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Prompt | Text | The text prompt is processed through various visual information extraction models (Canny, Depth, Redux), each uniquely interpreting different features of the input image to generate new visuals. | | Style | Different kinds of stylistic presets (*3D, Anime, Cyberpunk, Extreme Detailer, Game Asset, Logo, Midjourney, Motion Blur, Panorama, Pixar, Pixel, Retrofuturism, Studio Ghibli, Tarot Card, Thermal Image, Vector, Victorian Drawing, Watercolor*). | The style presets modify the rendering phase by influencing color schemes, textures, and artistic techniques, transforming the structural output from ControlNet into diverse visual interpretations. | | Strength | 0%-100% | The strength parameter alters the amount that the text influences the output image versus the source image. The higher the strength is, the more influence the text has. | | Image Size |

Auto,

Landscape - 16:9 (576x1024),

Portrait - 4:3 (768x1024),

1:1,

Landscape - 4:3 (1024x768), Portrait - 9:16 (1024x576).

| This changes the output size of the node with a couple of options. | | Seed | Seed | The seed is a deterministic number that indexes generations from the model. It's typically randomized, but you can set a seed if there's a particular output you're looking for! Keep in mind that all parameters must be the same in order for a given seed's output to persist. | # Video Node Learn the basics of video nodes The video node enables dynamic storytelling, temporal exploration, emotional resonance, and immersive narrative construction. It bridges movement and sound, allowing you to craft experiences that evolve over time, engage viewers emotionally, and convey complex ideas with fluid transitions. Video nodes provide a canvas for experimenting with pacing, rhythm, and atmosphere, enabling layered storytelling that unfolds with intention. They invite interactivity and viewer interpretation, enhancing narrative depth through motion, soundscapes, and sequential imagery. Here is a quick introduction on how to get started with video nodes: {% embed url="" %} ## Models Visit our [Video Models](/models/video-models) section to learn about the video models and capabilities available in the Video Node. # Text to Video ## Summary Text-to-Video models generate videos by taking a text prompt as input, using the prompt to guide scene composition, motion, and style, resulting in dynamic video content that matches the described narrative or visual concept.
### Parameters These are parameters that are applicable to all our base models.
ParameterTypeEffect on Output
PromptTextThe text prompt is processed through the selected model.
Aspect Ratio

Landscape - 16:9 (576x1024),

Portrait - 9:16 (1024x576), Square - 1:1.

This changes the output size of the node with a couple of options.
SeedSeedThe seed is a deterministic number that indexes generations from the model. It's typically randomized, but you can set a seed if there's a particular output you're looking for! Keep in mind that all parameters must be the same in order for a given seed's output to persist.
### How to use Prompting is everything for Text to Video. Refer to the [Text to Text](/nodes/text-node/text-to-text) card to explore more workflow combinations, or browse example workflows on the [Community page](https://app.flora.ai/community). # Image(s) to Video ## Summary Image-to-Video models generate videos by taking a single image or multiple images as input, using the input image(s) to guide scene continuity, motion, and transitions, resulting in dynamic video content that expands the static visuals into a moving narrative while maintaining visual consistency.
### Parameters These are parameters that are applicable to all our base models.
ParameterTypeEffect on Output
PromptTextThe text prompt is processed through the selected model.
SeedSeedThe seed is a deterministic number that indexes generations from the model. It's typically randomized, but you can set a seed if there's a particular output you're looking for! Keep in mind that all parameters must be the same in order for a given seed's output to persist.
### Images to Video The Images to Video node generates a video by connecting multiple frames. You can input up to 9 frames and the model will fill in the gaps! It primarily uses IP-Adapter based morphing techniques that's happening in the backend. You can interact with it in a number of ways: * Input images by clicking the upload button in the node itself * Connect any image output to the node and it'll populate the images section * Reorder or delete images once populated * Use the prompt to help guide the output! # Audio to Video Audio-to-Video workflows turn the output of an audio node into an input for a video node, so the sound you've generated on the canvas: a voiceover, a line of dialogue, a sound effect, a music cue drives the visuals you produce next. This is how speech becomes a lipsynced performance, how a still image becomes a talking avatar, and how generated sound lands on the same canvas as the video it belongs to, without ever leaving FLORA. The shape of the workflow is always the same. You generate or upload audio in an audio node, connect that audio node's output to a video node, connect a second input (either a video of a speaker for lipsync or a reference image for avatars), and pick a model that accepts audio as an input. The video node treats the audio as a conditioning signal matching mouth movement to speech, aligning motion to rhythm, or syncing timing to a beat. Then is renders a clip that already has the sound baked in.
### Parameters
ParameterTypeEffect on Output
AudioAudioThe audio output from an audio node. Connect a generated voiceover, SFX, music track, or uploaded clip — this is the track that drives the video.
VideoVideoThe visual source the audio is applied to. Connect a video node for lipsync workflows, or an image node for talking-avatar generations.
PromptTextOptional text guidance passed to models that accept it alongside audio and visual inputs. Useful for steering delivery, style, or camera behavior on models that expose those controls.
DurationDerivedMost audio-to-video models produce a clip whose length matches the input audio. Trim the audio track before generating if you want a shorter output.
Audio to Video The Audio to Video workflow takes an audio track and generates a synced video from it. When you connect an audio node to a video node and switch to an audio-aware model. The model then reads the audio as part of the conditioning taking into account the delivery, timing, and motion in the output all follow the track you provided. You can interact with it in a number of ways: * Connect any audio node output to the video node to populate the audio input * Optionally connect an image node as a subject or scene reference * Use the prompt to describe the setting, camera, and style the model should render * Switch the model to WAN 2.6 or WAN 2.7 for audio-driven generation * Adjust aspect ratio and resolution to match your delivery format How to use Prompting matters for Audio to Video too — your audio carries dialogue and timing, but the prompt carries the scene.
# Audio Node Learn the basics of audio nodes The Audio Node brings audio generation and transcription to the FLORA canvas. You can generate speech from text, create sound effects, and transcribe audio to text — all integrated into your node-based workflows. ## Overview Audio nodes support two directions of generation: * **Text to Audio** — Generate speech or sound effects from a text prompt * **Audio to Text** — Transcribe or analyze audio content into text These modes are determined automatically based on how you connect the audio node to other nodes on your canvas. *** ## Text to Audio ### How It Works Connect a text prompt or text node output to an audio node to generate speech or sound effects. The audio node will produce an audio file based on your prompt and selected model. ### Voice Selection Audio nodes include an inline **voice selector** in the floating controls and pre-output state. You can: * **Browse voices** — Open the voice dropdown to see all available voices for the selected model * **Preview voices** — Click the play button next to any voice to hear a sample before selecting * **Switch voices** — Change the voice at any time without leaving the node The voice selector reads options directly from the model's available parameters, so the voice list updates automatically when you switch models. ### Supported Models Audio generation is available through providers including **ElevenLabs** for text-to-speech and sound effects. Available voices and capabilities vary by model. *** ## Audio to Text ### How It Works Connect an audio node's output to a text node to enable audio-to-text transcription. FLORA automatically detects the `Audio to Text` mode and routes the audio through a speech-to-text model. ### Supported Models Audio-to-text transcription is supported by: * **ElevenLabs** — Speech-to-Text via the ElevenLabs STT endpoint * **Gemini** — Audio input is sent as inline data alongside your text prompt, enabling audio analysis and transcription ### Connecting Audio to Text Nodes 1. Add an audio node to your canvas with generated or uploaded audio 2. Drag a connection from the audio node's right output handle to a text node's input 3. The text node automatically switches to `Audio to Text` mode 4. Run the text node to transcribe or analyze the audio *** ## Credit Costs Audio generation costs vary by model and duration. Before generating, hover over the **generate button** to see a credit cost tooltip showing: * **Estimated credit cost** for the generation * **Your available credits** Credits are charged when the generation starts. If a generation fails, credits are automatically refunded. *** ## Tips * **Preview voices before committing** — Use the play buttons in the voice selector to find the right voice for your project before spending credits * **Chain audio into text** — Connect audio outputs to text nodes for transcription, then feed that text into image or video prompts for end-to-end multimedia workflows * **Check model capabilities** — Different audio models support different voices, languages, and output qualities. Check the model details for specifics. *** *Last updated: April 2026* The audio node brings voice, sound, and sonic atmosphere into your FLORA canvas. It closes the loop between what you see and what you hear, letting you generate voiceovers, sound effects, and transcriptions alongside the image and video work already happening on your canvas. No jumping out to a separate tool, no booking scratch VO, no silent concepts. Audio nodes turn sound into another first-class node you can pipe into the rest of your workflow. Here is a quick introduction on how to get started with audio nodes: {% embed url="" %} ## Capabilities * Text-to-speech. Type a script, choose a voice, and get a voiceover back. Play it on the canvas, export as MP3 or WAV. * Text-to-SFX. Describe a sound effect in plain language and generate it in place. * Text-to-Music. Describe a sonic environment in plain language and recieve a matching track. * Audio-to-text. Transcribe an audio clip into a text node for editing, captioning, or downstream prompting. * Lipsync. Pair an audio node with a video node to drive a lipsynced performance. * FAUNA-aware. FAUNA can create and chain audio nodes for you as part of a multi-step workflow. ## Models Visit our [Audio Models](/models/audio-models) section to learn about the video models and capabilities available in the Video Node. # Text to Audio Text-to-Audio generates sound from a text prompt, letting you create voiceovers, sound effects, and music tracks directly on the canvas You can start an audio workflow from a text node. Write a script, describe a sound, or sketch the music you want, then connect it to an audio node and pick a model. What comes back is playable audio on the canvas. Ready to export, layer under a video, or drive a lipsynced performance in a video node. Speech Text-to-speech turns a written script into a voiceover. Type what you want said, pick a voice from the preset library, and generate. Good for narration on explainer videos, scratch VO for concept pitches, or any moment where a voice needs to land on the canvas without a recording session. You can interact with it in a number of ways: * Type or paste your script into the prompt field * Select a voice from the dropdown, or use a recorded custom voice asset * Use expressive prompt tags (e.g. whisper, shout, happy) on models that support them to shape delivery * Connect the output to a video node with a lipsync model to drive a lipsynced performance SFX Text-to-SFX generates individual sound effects from a plain-language description. Describe the sound — "heavy wooden door slamming shut in a stone hallway" — and generate. Useful for ambient texture, hit sounds, foley, and anywhere you'd otherwise be digging through a stock library. Best practices: * Be specific about the source and environment (material, space, distance) * Generate a few variations with different seeds and pick the strongest * Layer multiple SFX audio nodes for richer atmospheres Music Text-to-music generates short music tracks from a text prompt describing genre, mood, instrumentation, or reference. Useful for background beds on social clips, pitch videos, and mood-setting on ad concepts where you don't want to license a track. You can interact with it in a number of ways: * Describe the track — genre, tempo, instruments, mood, references * Generate variations across seeds to find the right feel * Pair music output with a video node for a scored clip How to use: Here are some example workflows using Text to Audio: * Write a script → generate speech → connect to a video node with a lipsync model → export a lipsynced clip * Generate a product-demo voiceover → layer a music bed from text-to-music → export a finished promo * Generate three SFX variations for a scene → pick the strongest → layer onto a generated video Models Visit our [Audio Models](/models/audio-models) section to learn about the speech, SFX, and music models available in the Audio Node. ## Limits & formats **Character limit (ElevenLabs Multilingual v2)** ElevenLabs Multilingual v2 has a 450-character limit per generation. Text longer than 450 characters is truncated — this is why a script with 800+ characters may only produce audio for the first \~450. To narrate longer scripts, split the text across multiple audio nodes and chain them. **Output format** Audio is delivered as MP3. WAV output is not currently available for ElevenLabs voice models in FLORA. **Bitrate** The MP3 bitrate is fixed at 128 kbps and cannot be changed. # Document Node Learn the basics of document nodes The Document Node brings PDFs onto your FLORA canvas as a first-class citizen. Drop in a research paper, a brand book, a script, or a slide deck — every page becomes browsable and ready to feed the rest of your workflow. ## Capabilities * **Drop a PDF, get a node.** Drag any PDF onto the canvas (or pick one from the Add menu → Upload). FLORA renders the first page instantly via an in-browser preview while the document uploads in the background. * **Page-by-page navigation.** Multi-page PDFs gain a thumbnail carousel below the node. Click any page to jump to it, or use the arrow keys when the node is selected. * **Full-screen reader.** Open the document in a focused viewer to read full pages, select text, and right-click any selection to spawn it as a new text node. * **Document and image outputs.** Connect a document node to a text node to pass the full document as context, or to an image node to use the currently-open page as visual input. * **Drag-out pages.** Drag any page thumbnail from the carousel onto the canvas to extract it as a new single-page document node — useful for isolating one figure or one slide and wiring it independently. * **Sharing and library.** Document nodes participate in share links, view-only sessions, and the Asset Library just like any other media node. ## Connections * **Text nodes** — the full document is passed as context to the downstream model. * **Image nodes** — the currently-open page is used as visual input. Navigate pages on the document node to change which page is sent downstream. * **Batch nodes** — process every page in parallel through whatever pipeline you wire downstream. {% hint style="info" %} Even more connectivity is coming soon! If you have thoughts, comments, or suggestions, please let us know. {% endhint %} ## See also * [PDF to Text](/nodes/document-node/pdf-to-text) — wiring documents into text workflows. * [PDF to Image](/nodes/document-node/pdf-to-image) — using the current page as image input. # PDF to Text ## Summary There are two ways to get text from a document node into your workflow: wire it to a downstream text node to pass the full document as context, or open the fullscreen reader and select specific text to spawn it as a text node. ## How to use ### Connect downstream 1. Drop a PDF onto the canvas to create a Document Node. 2. Drag a connection from the document node to a Text Node. 3. The downstream node receives the full document as context. ### Extract selected text from fullscreen 1. Click the fullscreen button on the document node to open the focused viewer. 2. Select any text on the page. 3. Right-click the selection and choose **Create text node** — FLORA spawns a text node on the canvas containing exactly what you selected. ## Example workflows * **Research summarizer** — Document Node → Text Node prompted with "Summarize the methodology of this paper in three sentences." * **Brand-voice rewrite** — Document Node (style guide) → Text Node ("Rewrite this draft in the voice and tone described above.") * **Pull a specific quote** — open fullscreen, select the passage, right-click → Create text node, then wire that text node wherever you need it. * **Script breakdown** — Document Node (screenplay) → Batch Node → per-scene Text Node prompted with "Pull out every prop and wardrobe note in this scene." {% hint style="info" %} Even more connectivity is coming soon! If you have thoughts, comments, or suggestions, please let us know. {% endhint %} # PDF to Image ## Summary Each page of a Document Node is rendered as an image. Connect a document node to an image node and the currently-open page is passed as visual input — navigate pages on the document node to change what gets sent downstream. ## How to use ### Connect downstream 1. Drop a PDF onto the canvas. 2. Use the page carousel (or arrow keys) to open the page you want to use. 3. Drag a connection from the document node to an Image Node 4. The downstream node receives the currently-open page as visual context. Switch pages on the document node to swap which page is sent. ### Drag a page out 1. Hover the page carousel under the document node. 2. Grab any thumbnail and drag it onto an empty area of the canvas. 3. FLORA extracts that page into a new single-page PDF and creates a new Document Node pointing to it — useful when you want to operate on one figure or one slide independently of the rest of the document. ## Example workflows * **Reference-image generation** — Document Node (mood board PDF) → Image Node prompted with "Generate a frame that matches this composition." {% hint style="info" %} Even more connectivity is coming soon! If you have thoughts, comments, or suggestions, please let us know. {% endhint %} # Layer Editor Compose multiple images into a single layered composition The Layer Editor is a powerful compositing tool that lets you combine multiple images into a single layered composition. It provides a canvas-based editing environment similar to professional design tools, enabling precise control over positioning, sizing, and layering of visual elements. {% embed url="" %} ## Overview The Layer Editor transforms a node into a full-featured composition workspace where you can: * **Combine multiple images** from different sources into one composition * **Precisely position and transform** each layer independently * **Control layer ordering** to create depth and visual hierarchy * **Export the final composition** as a single unified image This is ideal for creating collages, mood boards, social media graphics, or any project that requires combining multiple visual elements. *** ## Getting Started ### Adding a Layer Editor Node {% stepper %} {% step %} **Add the Node** From the node menu, select **Layer Editor** to add it to your canvas. {% endstep %} {% step %} **Connect Source Images** Connect image nodes, video frames, or other visual sources to the Layer Editor. Each connected source becomes a layer in your composition. {% endstep %} {% step %} **Enter Edit Mode** **Double-click** the Layer Editor node, or click **Layer Editor** in the node toolbar to activate the editing interface. The canvas becomes interactive, displaying all your layers and editing controls. {% endstep %} {% endstepper %} *** ## Working with Layers ### The Layers Panel The Layers Panel on the side of the editor displays all layers in your composition. Layers are listed from top to bottom, with the topmost layer in the list appearing in front of other layers on the canvas. **Layer controls include:** * **Visibility toggle** - Show or hide individual layers * **Lock toggle** - Prevent accidental edits to a layer * **Layer name** - Click to rename for better organization * **Reorder** - Drag layers up or down to change their stacking order ### Selecting Layers * **Single click** on a layer in the canvas or panel to select it * **Ctrl/Cmd + Click** to toggle selection and select multiple layers * **Click empty canvas space** to deselect all layers *** ## Transforming Layers ### Moving Layers * **Drag** any selected layer to reposition it on the canvas * **Arrow keys** nudge the layer by 1 pixel * **Shift + Arrow keys** nudge by 10 pixels for faster movement ### Resizing Layers Click a layer to reveal the transform handles: * **Corner handles** - Resize while maintaining aspect ratio * **Edge handles** - Resize in one direction only * **Double-click** a layer to automatically fit it to the canvas frame {% hint style="info" %} Hold **Shift** while resizing to toggle aspect ratio lock on or off. {% endhint %} ### Rotating Layers Hover near a corner handle to see the rotation cursor. Click and drag to rotate the layer freely. ### Flipping Layers Use the transform actions panel to: * **Flip Horizontal** - Mirror the layer left-to-right * **Flip Vertical** - Mirror the layer top-to-bottom *** ## Layer Alignment The alignment toolbar provides quick alignment options for selected layers: | Button | Action | | ------------ | --------------------------------------------- | | Align Left | Align layer to the left edge of the canvas | | Align Right | Align layer to the right edge of the canvas | | Center H | Center layer horizontally on the canvas | | Align Top | Align layer to the top edge of the canvas | | Align Bottom | Align layer to the bottom edge of the canvas | | Center V | Center layer vertically on the canvas | | Center All | Center layer both horizontally and vertically | *** ## Layer Ordering Control which layers appear in front of others: ### Using the Context Menu Right-click a layer to access ordering options: * **Send to Top** - Move layer to the front of all others * **Send to Bottom** - Move layer behind all others * **Send Up** - Move layer one position forward * **Send Down** - Move layer one position backward ### Keyboard Shortcuts | Action | Shortcut | | --------- | -------------- | | Send Up | `Cmd/Ctrl + ]` | | Send Down | `Cmd/Ctrl + [` | ### Drag Reordering In the Layers Panel, drag layers up or down to change their stacking order visually. *** ## Appearance Controls ### Opacity Adjust layer transparency using the opacity slider in the Appearance Panel: * **100%** - Fully opaque (default) * **0%** - Fully transparent Changes preview in real-time as you adjust the slider. *** ## Canvas Settings ### Frame Size The Layer Editor uses a compositor frame that defines the output dimensions: * **Default size**: 2048 × 2048 pixels * **Maximum size**: 4096 × 4096 pixels * **Minimum size**: 100 × 100 pixels ### Snap to Grid Enable snap-to-grid for precise alignment. When enabled, layers snap to nearby edges and center points as you move them. *** ## Keyboard Shortcuts | Action | Shortcut | | --------------------- | ----------------------- | | Delete selected layer | `Delete` or `Backspace` | | Undo | `Cmd/Ctrl + Z` | | Redo | `Cmd/Ctrl + Shift + Z` | | Nudge 1px | Arrow keys | | Nudge 10px | `Shift + Arrow keys` | | Send layer up | `Cmd/Ctrl + ]` | | Send layer down | `Cmd/Ctrl + [` | *** ## Generating Output When you're satisfied with your composition: 1. All visible layers are automatically composited together 2. The Layer Editor generates a preview image 3. Connect the Layer Editor output to other nodes for further processing {% hint style="info" %} The compositor combines only **visible** layers. Toggle layer visibility to exclude layers from the final output without deleting them. {% endhint %} *** ## Tips for Better Compositions * **Lock finished layers** to prevent accidentally moving them while working on others * **Name your layers** descriptively to stay organized in complex compositions * **Use alignment tools** for professional, balanced layouts * **Work with the snap-to-grid** feature enabled for precise positioning * **Undo liberally** - the Layer Editor maintains a history of up to 50 changes *** ## Common Use Cases * **Social media graphics** - Combine multiple product photos into polished layouts * **Mood boards** - Arrange multiple reference images in one composition * **Collages** - Create artistic arrangements of generated images * **Presentations** - Layer diagrams and screenshots together * **Before/after comparisons** - Position related images side by side # Action Node Traditional editing tools that plug directly into your canvas workflows Action Nodes bring traditional editing tools—color grading, video trimming, frame extraction, text manipulation, and more—into your canvas as workflow steps. Instead of exporting media to an external editor and re-importing the result, you can drop an Action Node right into your workflow and connect it like any other node. {% embed url="" %} ## Overview Actions handle the kinds of operations you'd normally reach for a dedicated editing app to do: * **Adjusting color and tone** on an image or video * **Splitting or stitching** video clips together * **Extracting frames** from a video as still images * **Changing aspect ratios** with crop or pad options * **Manipulating text** with find-and-replace, splitting, or concatenation * **Splitting or merging audio tracks** from video files Because they live on the canvas alongside your generation nodes, Techniques, and Batch Nodes, actions slot naturally into any workflow—no round-tripping required. {% hint style="info" %} Need a tool that isn't in the catalog? Describe it in plain language and FLORA builds it for you—see [Custom Actions](/nodes/custom-actions). {% endhint %} *** ## Getting Started ### Adding an Action Node {% stepper %} {% step %} **Open the Add Node Menu** Press the **+** button or use the canvas shortcut to open the node menu. Under the **Utilities** section, click **Action**. {% endstep %} {% step %} **Choose an Action** A picker panel opens showing all available actions. Browse the list or use the search bar to find what you need. Each action shows its input and output types (image, video, text) so you know what to connect. {% endstep %} {% step %} **Connect Inputs** Drag a connection from an upstream node (an Image, Video, or Text node) to the Action Node's input handle. The action will process whatever media is connected. {% endstep %} {% step %} **Configure Parameters** Each action has its own set of adjustable parameters (sliders, dropdowns, color pickers, toggles). Adjust these to dial in the exact result you want. {% endstep %} {% step %} **Run** Click the **Run** button on the Action Node. The result appears in one or more output slots depending on the action. {% endstep %} {% endstepper %} *** ## Available Actions Actions span image, video, and text operations. Here's the full catalog: ### Image Actions | Action | What It Does | Key Parameters | | --------------------- | ------------------------------------------------------------------------------------------------------- | --------------------------------------------- | | **Color Grade Image** | Cinematic color grading with adjustable tone | Warmth, Contrast, Saturation, Brightness | | **Color Filter** | Apply preset filters: Grayscale, Sepia, Duotone, Clarendon, Moon, Nashville, Noir, Fade, and more | Filter preset, Intensity | | **Color Tint** | Tint an image with a chosen color using blend modes (Multiply, Screen, Overlay, Soft Light) | Tint Color, Intensity, Blend Mode | | **Filter Color** | Isolate, replace, or remove pixels matching a target color (chroma key, background removal, color swap) | Target Color, Tolerance, Mode | | **Change Image AR** | Change aspect ratio by center-cropping or padding without upscaling | Aspect Ratio, Fit (Crop/Pad), Background Mode | | **Rotate Image** | Rotate by any angle with optional canvas expansion | Angle, Expand Canvas, Background Color | | **Flip Image** | Mirror horizontally, vertically, or both | Direction | | **Blur** | Gaussian, Box, Motion, Radial, Bilateral, Bokeh, Tilt-Shift, or Target Color blur | Blur Type, Radius/Amount | | **Duplicate Image** | Output N copies of an image for branching into parallel workflow lanes | Number of Copies | ### Video Actions | Action | What It Does | Key Parameters | | ---------------------------- | ---------------------------------------------------------------------------------------------- | ----------------------------------------------------- | | **Color Grade Video** | Cinematic color grading for video | Contrast, Saturation, Brightness, Gamma | | **Video Color Filter** | Preset color filters for video: Grayscale, Sepia, Invert, Clarendon, Moon, Nashville, and more | Filter preset | | **Video Effect** | Stylistic effects: Vignette, Film Grain, Pixelate, Camera Shake, Chromatic Aberration, VHS | Effect type, Intensity | | **Stitch Videos** | Join multiple clips into one with optional transitions and aspect-ratio normalization | Transition type, Duration, Aspect Ratio, Fit Mode | | **Split Video** | Cut a video into segments by equal parts, fixed duration, or scene detection | Split Mode, Segment count, Sensitivity | | **Extract Video Frames** | Pull still frames from video as PNG images | Mode (Single, Evenly Spaced, Interval, Scene Changes) | | **Image to Video Ken Burns** | Animate a still image with smooth pan/zoom into a video | Duration, Zoom Level, Pan Direction, Easing | | **Video to Frame Grid** | Arrange extracted frames in a grid contact sheet | Rows, Columns, Cell Width, Gap | | **Video to Long Exposure** | Stack all frames into a single long-exposure image (Average, Lighten, Darken blend) | Blend Mode | | **Boomerang** | Play a video forward then reverse for a looping clip | Loops, Speed, Frame Stride | | **Reverse Video** | Play a video backwards | Reverse Audio, Keep Original Audio | | **Speed Up Video** | Increase playback speed | Speed Factor, Keep Audio | | **Slow Down Video** | Decrease playback speed | Slow Factor, Keep Audio | | **Watermark** | Burn a text or image watermark into a video | Text/Image, Position, Opacity, Scale | | **Greenscreen Remove** | Chroma-key out a background color and output transparent video | Color Preset, Similarity, Edge Blend | | **Duplicate Video** | Output N copies of a video for branching workflows | Number of Copies | ### Text Actions | Action | What It Does | Key Parameters | | -------------------- | ------------------------------------------------------------------------------------ | ------------------------------------------- | | **Split Text** | Split text into parts by separator, regex, paragraph, line count, or character count | Split Mode, Separator, Max Parts | | **Find and Replace** | Find and replace text with optional regex, case-insensitive, and whole-word modes | Find, Replace, Use Regex, Case Sensitive | | **Concat Text** | Combine multiple text inputs into one | Separator, Prefix, Suffix, Trim, Skip Empty | *** ## How Actions Work ### Inputs and Outputs Every action declares typed **input slots** and **output slots**. The Action Node shows modality badges (image, video, text) on each slot so you can see at a glance what connects where. * **Single inputs** accept exactly one connection (e.g., one image to color-grade) * **Dynamic inputs** accept multiple connections (e.g., Stitch Videos takes 2+ video clips) * **Single outputs** produce one result (e.g., a color-graded image) * **Dynamic outputs** produce a variable number of results (e.g., Split Video outputs multiple video segments) ### Parameters Actions expose configurable parameters directly on the node—no prompt writing needed. Parameter types include: * **Sliders** for numeric values (warmth, contrast, speed factor) * **Dropdowns** for preset options (blur type, filter style, aspect ratio) * **Color pickers** for color-based operations (tint color, background color) * **Toggles** for boolean options (keep audio, expand canvas) * **Text fields** for string inputs (watermark text, separator) Some parameters are conditional—they only appear when a related setting is active. For example, the Blur Amount slider in Change Image AR only shows when Fit is set to "Pad" and Background Mode is "Blur." ## Actions in Workflows Actions become especially powerful when connected into larger workflows alongside other Flora nodes. ### Chaining with Generation Nodes Use actions to prepare media before or after AI generation: ``` [Text Node] → [Image Node: Generate] → [Color Grade Image] → Final Output "product on white bg" warmth=0.2, contrast=1.3 ``` Generate an image with AI, then apply deterministic color grading for a consistent brand look. ### Combining with Techniques Actions work seamlessly as steps before or after Techniques: ``` [Image Upload] → [Change Image AR] → [Technique: Product Lifestyle] → Final Output 9:16 portrait ``` Reformat your image to the right aspect ratio before feeding it into a multi-step technique. ### Actions Inside Techniques Action Nodes can be included directly inside a Technique's workflow graph. When you build a technique in the Technique Builder, you can wire action nodes alongside generation nodes—the action executes via the sandbox during the technique run, and its outputs flow through to downstream nodes or technique outputs. This means you can package deterministic post-processing (color grading, frame extraction, audio splitting, etc.) as part of a reusable technique pipeline. For example: ``` [Input Image] → [Image Node: Generate styled version] → [Color Grade Image] → [Output] ``` The action node runs automatically as part of the technique—users of the technique don't need to set up or configure the action separately. {% hint style="info" %} When viewing a technique's internal workflow (via **View workflow** on a Technique Node), action nodes appear as violet "Action" blocks with a distinct icon, making them easy to distinguish from generation nodes. {% endhint %} ### Batch Processing Connect a [Batch Node](/nodes/batch-node) to an Action Node to apply the same operation across many items at once: ``` [Image 1] ─┐ [Image 2] ─┼─→ [Batch Node] ─→ [Color Grade Image] → 4 graded outputs [Image 3] ─┤ (4 items) warmth=0.15 [Image 4] ─┘ ``` Every item in the batch is processed with identical settings—perfect for maintaining visual consistency across a set of assets. ### Multi-Step Action Chains Actions can chain together for complex transformations without any AI: ``` [Video Upload] → [Split Video] → [Extract Frames] → [Color Filter: Noir] 3 segments per segment per frame ``` ### Cross-Modality Conversions Some actions convert between modalities, enabling creative cross-format workflows: ``` [Image] → [Ken Burns Video] → [Stitch Videos] → Final Video [Image] → [Ken Burns Video] ─┘ ``` Turn still images into animated clips and stitch them into a sequence—all without AI generation. ### Audio Workflows The audio actions enable creative audio-visual workflows: ``` [Video Upload] → [Split Audio from Video] → Audio track + Muted video ``` ``` [Video] ─→ [Merge Audio into Video] → Video with new soundtrack [Audio] ─┘ ``` Extract audio for transcription or remixing, or replace a video's audio track with a voiceover or music track generated from an Audio Node. *** ## Tips * **Test with one input first** before connecting a batch—dial in your parameters on a single item, then scale up * **Chain actions freely**—they're lightweight and fast, so stacking multiple actions adds minimal time * **Use Duplicate Image/Video** to branch a single input into parallel workflow lanes with different downstream processing * **Dynamic outputs** (like Split Video or Extract Frames) produce multiple result nodes—each can feed into its own downstream workflow * **Audio actions stream-copy** when source codecs are mp4-native (e.g., h264 + AAC), making them near-instant for common video formats # Custom Actions Describe the tool you wish existed, and FLORA builds it for your canvas The [Action Node](/nodes/action-node) catalog covers the most common editing operations—but sometimes you need a tool that doesn't exist yet. Custom Actions let you describe the tool you want in plain language, and FLORA builds it for you on the spot. No plugins, no external apps, no waiting for a feature request. Type "add a thin white border and my caption to the bottom of this image," and seconds later you have a working, reusable node with its own controls—sliders, color pickers, dropdowns—sitting on your canvas like any other action. *** ## Creating a Custom Action {% stepper %} {% step %} **Add the node** Open the Add Node menu with the **+** button and choose **Generate Action** ("Build with AI from a prompt"). A new node appears with a single text field: *Describe an action...* {% endstep %} {% step %} **Describe what you want** Write what the tool should do, like you'd explain it to a colleague: "split this image into a 3×3 grid," "convert this video to a looping GIF," "turn each line of this text into a numbered list." If you've already connected inputs, FLORA sees their types (image, video, audio, or text) and builds the action to match. {% endstep %} {% step %} **Press Enter** FLORA generates the action in a few seconds. The node updates with a name, input slots, and any controls the tool needs—dial them in just like a prebuilt action. {% endstep %} {% step %} **Connect and Run** Wire up your inputs, adjust the controls, and click **Run**. Results appear as output nodes, ready to feed into the rest of your workflow. {% endstep %} {% endstepper %} ### Refining and fixing Custom actions are conversational, not one-shot: * **Edit** — click the **Edit** pill on the node to describe a change ("make the border thicker," "add a slider for opacity") and FLORA updates the action in place. * **Troubleshoot** — if a run fails, a **Troubleshoot** button appears with the error. One click sends the details to [FAUNA](/editor/fauna), which diagnoses and fixes the action for you. *** ## What Custom Actions Can Do Custom actions work with all four canvas modalities—**image, video, audio, and text**—and can accept multiple inputs, produce multiple outputs, and convert between formats. They run in a secure sandbox with a rich toolkit for image manipulation, video processing, and text handling built in. ### Cool things people build * **Stylized effects models don't do well** — halftone dots, dithering, ASCII art, scan lines, precise duotones with your exact hex values. * **Brand enforcement tools** — apply your exact brand colors, margins, watermark, and caption style to any image. Deterministic and pixel-perfect, unlike a prompt. * **Contact sheets and grids** — tile a set of images into a labeled grid for client review, or slice one image into a 9-tile Instagram grid. * **Format converters** — video to looping GIF, image set to sprite sheet, audio waveform to image. * **QR codes and overlays** — generate a QR code from a text node and composite it onto a poster. * **Prompt machinery** — split a CSV of product names into individual prompts, wrap each in your house style, and feed them to a [Batch Node](/nodes/batch-node). * **Video utilities** — extract a thumbnail at an exact timestamp, burn a countdown timer into a clip, assemble stills into a timelapse. The pattern: whenever you need something **precise, repeatable, or mechanical**—the kind of task where an AI model's creative interpretation is a bug, not a feature—a custom action is the right tool. *** ## Custom Actions in Workflows Custom actions behave like any other node, so they compose with everything on the canvas: * **Chain with generation nodes** — generate an image with AI, then run your custom "brand kit" action for a consistent finish. * **Batch them** — connect a [Batch Node](/nodes/batch-node) and the Run button becomes **Run N**, processing every item with identical settings. * **Use them in Techniques** — custom actions can be steps inside a [Technique](/nodes/techniques), mixing deterministic processing with AI generation in one repeatable workflow. ``` [CSV of 50 product names] → [Custom Action: format prompts] → [Batch Node] → [Image Node] → [Custom Action: brand kit] → 50 on-brand assets ``` *** ## Good to Know * **Be specific about inputs and outputs.** "Take an image and a text caption, output one image" generates a better tool than "make a caption thing." * **Start simple, then iterate.** Get a basic version running, then use **Edit** to layer on options and controls. * **Actions are project-scoped.** A custom action lives in the project where you created it; recreate it elsewhere by describing it again. * **Runs take up to 5 minutes.** Long video operations that exceed the limit are better split into steps. * **Custom actions require an eligible plan.** If you see "Upgrade to generate custom actions" in the node menu, check your [workspace plan](/plans-and-billing/pricing). # Batch Node Process multiple items through the same generation workflow The Batch Node allows you to aggregate multiple images, videos, or text items and process them all through one or more downstream generation nodes. Instead of manually running the same operation multiple times, the Batch Node automates bulk processing—saving time and ensuring consistency across all your outputs. {% embed url="" %} ## Overview The Batch Node is designed for workflows where you need to apply the same transformation to multiple inputs: * **Upscale a collection of images** using the same enhancement model * **Generate variations** from multiple reference images * **Apply consistent styling** across a set of base images * **Process multiple text prompts** through the same generation pipeline Think of it as a "for each" loop for your creative workflow—define the operation once, and the Batch Node applies it to every item in the collection. {% embed url="" %} *** ## Getting Started ### Adding a Batch Node {% stepper %} {% step %} **Add the Node** From the node menu, select **Batch** to add it to your canvas. {% endstep %} {% step %} **Add Items to the Batch** Populate your batch with items using any of these methods: * **Connect upstream nodes** - Link image, video, or text nodes to the batch input * **Upload files** - Click the Upload button or drag-and-drop files onto the node * **Drag from asset library** - Pull items from your workspace history or assets {% endstep %} {% step %} **Connect a Generation Node** Link the Batch Node output to a generation node (Image, Video, or Text). This defines what operation will be applied to each item. {% endstep %} {% step %} **Run the Generation** Execute the downstream node. The batch automatically processes all items in parallel, creating separate outputs for each input. {% endstep %} {% endstepper %} *** ## Adding Items to a Batch ### Connecting Upstream Nodes The most common way to populate a batch is by connecting other nodes: 1. Create your source nodes (image generators, uploads, etc.) 2. Connect each source to the Batch Node's input handle 3. When source nodes produce output, those outputs appear as items in your batch Multiple nodes can feed into the same batch, allowing you to collect outputs from various sources. ### Uploading Files Click the **Upload** button in the Batch Node to select files from your computer, or simply drag-and-drop files directly onto the node. **Supported file types:** | Modality | Formats | | -------- | ------------------------- | | Images | PNG, JPG, JPEG, WEBP, SVG | | Videos | MP4, WEBM, MOV | | Text | CSV | ### Uploading a CSV You can upload a `.csv` file to quickly populate a text batch. FLORA reads the first column of each row and creates one batch item per row. All rows are included (there is no header skip), so remove any header row you don't want processed. Drag-and-drop a CSV onto the node, or click **Upload** and select a `.csv` file. You'll see a toast confirming how many rows were added. If the batch already has items, the new rows are appended up to the 100-item limit. ### Dragging from Assets Drag items from the workspace sidebar directly onto the Batch Node: * **History** - Previously generated images or videos * **Assets** - Saved items in your workspace * **Unsplash** - Stock photos from the Unsplash integration ### Splitting Text Inputs When text nodes are connected to a batch, a **text split toolbar** appears above the node. This lets you split each connected text node's content into separate batch items using a delimiter. The toolbar offers three modes via the **Input split** dropdown: * **None** (default) — Each connected text node is one batch item * **By line** — Splits on newlines, so each line becomes its own batch item * **Custom** — Splits on any string you type (e.g. `;`, `|`, `//`) For example, if a text node contains `"cat\ndog\nbird"` and you select **By line**, the batch will contain three items: "cat", "dog", and "bird". Empty and whitespace-only segments are automatically filtered out. {% hint style="info" %} Text splitting and CSV upload work together — CSV-uploaded items are preserved when you change the split mode, and vice versa. {% endhint %} *** ## Modality Locking The Batch Node locks to a single content type (modality) once you add your first item: * Add an image → the batch becomes an **image batch** * Add a video → the batch becomes a **video batch** * Connect a text node → the batch becomes a **text batch** {% hint style="info" %} You cannot mix different content types in the same batch. All items must be the same modality (all images, all videos, or all text). {% endhint %} If you remove all items from a batch, the modality lock is released, allowing you to start fresh with a different content type. *** ## The Batch Interface ### Header Displays the batch icon, the current modality (if set), and the total item count. ### Item List Shows all items currently in the batch with: * **Thumbnail** for images and videos * **Text preview** (first 100 characters) for text items * **Source label** indicating where the item came from * **Delete option** to remove individual items ### Footer * **Upload button** to add more items * **Add Generation Node** button to quickly connect a downstream processing node *** ## Processing a Batch ### How Batch Execution Works When you run a generation node connected to a batch: 1. **Usage validation** - FLORA checks you have sufficient budget for all items 2. **Task creation** - A separate generation task is created for each batch item 3. **Parallel processing** - All items are processed simultaneously 4. **Output creation** - Each item produces its own output in the generation history ### Cost Calculation Batch processing costs are calculated as: ``` Total Cost = Generation Cost per Item × Number of Items ``` For example, if upscaling costs $0.05 per image and your batch contains 10 images, the total cost is $0.50. {% hint style="info" %} FLORA validates your usage budget before starting. If you don't have enough budget for all items, the batch won't start. {% endhint %} ### Generation Limits * **Maximum items per batch execution**: 100 items * If your batch contains more than 100 items, you'll need to run the generation multiple times *** ## Connecting to Generation Nodes ### Quick Connect Click the **arrow button** in the Batch Node footer to open a node selection modal. Choose a generation node type to automatically create and connect it. ### Manual Connection Drag from the Batch Node's output handle to any compatible downstream node: * **Image Node** - For image-to-image transformations, upscaling, style transfer * **Video Node** - For image-to-video generation * **Text Node** - For image-to-text or text processing * **Technique Node** - For running a full multi-step technique pipeline on each batch item ### How Connections Work When a batch is connected to a downstream node: * **Image batches** replace the image input for each generation * **Video batches** replace the video input for each generation * **Text batches** inject text content into the prompt parameter The downstream node's settings (model, parameters, prompt) apply identically to every item in the batch. ### Multiple Downstream Nodes A single Batch Node can connect to multiple generation nodes downstream. Each downstream node independently processes every item in the batch, so one collection of inputs can feed into several different operations at once. {% embed url="" %} *** ## Combining Two Batches (Batch × Batch) You can connect **two** Batch Nodes to the same downstream generation node to produce results for every pairing — each output combines one item from Batch A with one item from Batch B. This is ideal for exploring style × subject grids, prompt × reference comparisons, or any "every-of-these against every-of-those" workflow. When a generation node has exactly two Batch Nodes connected, a **Zip / Cross** toggle appears on the node: * **Cross** (default) — Cartesian product. If Batch A has N items and Batch B has M items, the node generates N × M outputs, one for every possible pair. * **Zip** — Positional pairing. Pairs items by index, producing `min(N, M)` outputs. The toggle is only enabled when both batches have matching item counts; when sizes differ, the node stays in Cross mode automatically. Switching modes updates the generate button's item count and credit cost in real time. Your choice persists with the node. {% hint style="info" %} The Zip / Cross toggle supports all four block types — Image, Video, Text, and Audio. Connecting three or more Batch Nodes to a single generation node is not supported. {% endhint %} ### Tracking Which Output Came From Which Pair Every result is tagged with its source pair (one item from each parent batch), so you can navigate between inputs and outputs: * **Input → Output** — Click an item in either parent batch to scroll the result carousel to the generation that used that item. * **Output → Input** — Selecting a result highlights the matching item in both parent batches. This two-way sync works the same in Zip and Cross mode, and stays correct even if you later reorder items in either parent batch. ### Matrix View (Fullscreen) When a generation node is in **Cross** mode, the fullscreen view (double-click any result) adds a new **Matrix** layout option alongside Single and Masonry. The Matrix is a grid where rows correspond to Batch A items and columns to Batch B items. Every cell shows the generation produced by that (row, column) pair, and sticky row and column headers display each parent batch's thumbnails so you can always see the inputs. In the Matrix view you can: * **Hover a cell** to reveal the two source thumbnails and a download button * **Click a completed cell** to open that result in the Single view, with its metadata panel, prompt drawer, and keyboard navigation * **Download a cell** directly — same formats and filename scheme as the rest of fullscreen Pending cells show a spinner in place; failed cells show an error icon with a tooltip describing the failure. {% hint style="info" %} The Matrix option appears only in Cross mode when both parent batches have thumbnailable items. Zip-mode and single-batch results continue to use the Masonry and Single layouts. {% endhint %} ### Chaining Beyond Two Batches You are not limited to mixing two live batches at the same node. A Cross (or Zip) arm can also be the **history of a previous batch generation** — so you can feed a batch result into another block, then combine that block's history with a third batch to explore a new dimension. The Matrix view supports these combinations too: row or column headers draw from the history arm when that input is a prior generation rather than a live batch. *** ## Workflow Examples ### Example 1: Batch Upscaling ``` [Image 1] ─┐ [Image 2] ─┼─→ [Batch Node] ─→ [Image Node: Upscale] → 3 upscaled outputs [Image 3] ─┘ (3 items) (Magnific 2x) ``` All three images are upscaled using the same model and settings. ### Example 2: Style Transfer Across Multiple Images ``` [Photo 1] ─┐ [Photo 2] ─┼─→ [Batch Node] ─→ [Image Node] → 4 stylized outputs [Photo 3] ─┤ (4 items) "anime style" [Photo 4] ─┘ ``` Apply the same "anime style" prompt to all four photos. ### Example 3: Generate Videos from Multiple Images ``` [Product Shot 1] ─┐ [Product Shot 2] ─┼─→ [Batch Node] ─→ [Video Node] → 3 product videos [Product Shot 3] ─┘ (3 items) "slow zoom" ``` Create video animations from multiple product images. ### Example 4: Style × Subject Grid (Batch × Batch, Cross Mode) ``` [Photo 1] ─┐ [Photo 2] ─┼─→ [Batch A: 3 subjects] ─┐ [Photo 3] ─┘ │ ├─→ [Image Node, Cross] → 3 × 3 = 9 outputs [Anime] ─┐ │ (open fullscreen → Matrix) [Pixar] ─┼─→ [Batch B: 3 styles] ──┘ [Watercolor] ─┘ ``` Generate every combination of 3 photos across 3 styles, then compare them side-by-side in the Matrix view. ### Example 5: Batch Technique Processing ``` [Product 1] ─┐ [Product 2] ─┼─→ [Batch Node] ─→ [Technique: Lifestyle Shot] → 3 lifestyle images [Product 3] ─┘ (3 items) ``` Run an entire multi-step technique pipeline on each item in the batch. Each product image is processed through the full technique workflow independently, producing complete outputs for every item. *** ## Managing Batch Items ### Removing Items Hover over an item in the batch list and click the delete button to remove it. Removing an item also disconnects it from any source node. ### Clearing the Batch Remove all items to reset the batch completely, including its modality lock. ### Item Sources Each item displays its source: * **Node name** - For items from connected nodes * **"Uploaded file"** - For manually uploaded items *** ## Tips for Effective Batch Processing * **Consistent quality** - Ensure all source images have similar quality and dimensions for best results * **Test first** - Run a single item through your generation node before processing the full batch * **Check your budget** - Verify you have enough usage budget before starting large batches * **Monitor progress** - Each item generates independently, so you can see results as they complete * **Organize sources** - Name your source nodes clearly to track which outputs came from which inputs *** ## Limitations | Limitation | Details | | ------------------------ | ------------------------------------------------------------------------------------------------------------------ | | Single modality only | Cannot mix images, videos, and text in one batch | | CSV first column only | CSV upload reads only the first column per row | | 100 items per execution | Larger batches require multiple runs | | Same settings for all | Cannot customize parameters per item | | Two batches max per node | A single generation node accepts at most two Batch Nodes (Cross / Zip) — connecting a third disables batch × batch | | Zip requires equal sizes | Zip mode is only enabled when both parent batches have the same item count; otherwise the node stays in Cross mode | *** ## Using Batch Nodes Inside Techniques Batch nodes can be included inside Techniques, enabling fan-out processing as part of a reusable workflow. When a Technique contains a Batch node: * The batch fan-out executes as an internal step during the technique run * Collection outputs are surfaced on the technique node's output slots * Users can view each item in the collection from the sidebar, detail view, or App Mode This lets technique authors build workflows that process multiple items — for example, a technique that takes one product photo and generates a batch of ad variations from it. {% hint style="info" %} Chaining multiple batch nodes inside a technique in a way that creates explosive fan-out (e.g., batch → batch) is blocked at publish time to prevent runaway costs. {% endhint %} *** ## Common Use Cases * **Product photography** - Apply consistent editing to an entire product line * **Social media content** - Generate multiple variations for A/B testing * **Asset preparation** - Upscale or enhance a collection of images at once * **Video creation** - Turn a series of images into individual video clips * **Style exploration** - Apply different models to the same set of inputs (using multiple batches) * **Style × subject grids** - Use Batch × Batch in Cross mode to explore every combination of a prompt set against a reference set, then review results in the Matrix view * **A/B pairing** - Use Zip mode to pair two parallel lists (e.g. prompts with matching reference images) at the same index # Export Node Save and export your generated media to local downloads or Google Drive, and import files from Drive The Export Node is your output destination on the canvas. It lets you download generated images, videos, and audio in multiple formats, or push them directly to Google Drive -- all without leaving FLORA. ## Overview Every generation node on the canvas already has a download button in its toolbar, but the Export Node gives you a dedicated, persistent place to: * **Download media** in your preferred format (PNG, JPG, MP4, MOV, GIF, and more) * **Export to Google Drive** with a one-click connection to any folder in your Drive * **Export to Shopify** to publish visuals straight to your store's Files library—see [Export to Shopify](/more/export-to-shopify) * **Batch-export outputs** when multiple items are wired in Think of it as the final stop in your workflow -- wire any node's output into the Export Node to save or sync it externally. *** ## Getting Started ### Adding an Export Node {% stepper %} {% step %} **Open the Add Node Menu** Press the **+** button or use the canvas shortcut to open the node menu. Under the **Utilities** section, click **Export**. You can also drag from any node's output handle onto empty canvas — the Export node appears in the quick-add picker. {% endstep %} {% step %} **Place It on the Canvas** The Export Node appears as a gold-bordered block with an upload icon. It starts empty with the prompt: "Wire any node into here to save its outputs." {% endstep %} {% step %} **Connect an Upstream Node** Drag a connection from any generation node (Image, Video, Audio, or Text) to the Export Node. When that node produces output, it becomes available for export. {% endstep %} {% endstepper %} *** ## Downloading Media Every generation node on the canvas includes a **Download** button in its hover toolbar. Clicking it opens a format dropdown so you can choose the output format before downloading. ### Image Formats | Format | Description | | -------- | --------------------------------------------- | | **PNG** | Lossless format, best for sharp graphics | | **JPG** | Compressed format, smaller file size | | **WEBP** | Modern web format, good quality-to-size ratio | | **SVG** | Vector format (transcoded from raster via AI) | {% hint style="info" %} If the source image is already an SVG, SVG appears first in the format list. {% endhint %} ### Video Formats | Format | Description | | ---------------- | --------------------------------------------- | | **Original** | Download in the original generated format | | **MP4 (h264)** | Widely compatible, great for sharing | | **MP4 (h265)** | Better compression, smaller files | | **MOV (h264)** | Apple-compatible container with h264 encoding | | **MOV (ProRes)** | Professional editing format, high quality | | **GIF** | Animated image format, great for short loops | ### Audio Formats Audio nodes download in their original format. Supported formats include **MP3**, **M4A**, **WAV**, **AAC**, **OGG**, and **FLAC**. The download preserves whichever format the audio was generated or uploaded in. ### Keyboard Shortcuts When the download dropdown is open, press a number key (**1**, **2**, **3**, etc.) to quickly select the corresponding format without clicking. ### Bulk Download When a node has multiple items in its generation history, a **Download All** button appears in the toolbar. This bundles all history items into a single ZIP file. *** ## Exporting to Google Drive The Export Node supports direct export to Google Drive. Once connected, your generated media is uploaded to a chosen Drive folder automatically. ### Connecting Google Drive {% stepper %} {% step %} **Click "Connect to Google Drive"** On a new Export Node, click the gold **Connect to Google Drive** button. This initiates an OAuth sign-in flow. {% endstep %} {% step %} **Authorize FLORA** A Google sign-in window opens. Grant FLORA permission to access your Google Drive. {% endstep %} {% step %} **Pick a Destination Folder** After authorization, a Google Drive folder picker opens. Browse your Drive with proper folder navigation — folders appear grouped before files and can be traversed into. You can choose any folder from **My Drive** or from **Shared Drives** (accessible via a dedicated tab). Click **Select** to confirm. {% endstep %} {% step %} **Done** The Export Node updates to show the connected folder name. The node icon switches to the Google Drive logo. {% endstep %} {% endstepper %} ### Shared Drives Google Shared Drives (formerly Team Drives) are fully supported for both import and export. When browsing for a destination folder, switch to the **Shared Drives** tab to see all shared drives your Google account has access to. You can navigate into nested folders within a shared drive and select any folder as your export destination. ### Managing the Destination Once connected, hover over the Export Node to reveal the toolbar. Click the **folder dropdown** to: * **Change folder** -- Open the folder picker again to select a different destination * **Clear destination** -- Remove the Drive connection entirely The dropdown also displays the connected Google account email for reference. ### How Drive Export Works When media is exported to Google Drive: 1. FLORA streams the file directly from its CDN to your Drive folder 2. Multiple files are uploaded with bounded concurrency (up to 4 simultaneous uploads) 3. Each file is uploaded independently -- if one fails, the others still complete 4. A progress bar tracks real-time upload progress with byte-level accuracy 5. A toast notification shows progress and confirms when the export is complete, with a direct **Open in Drive** link ### Export Node Toolbar Hover over a connected Export Node to access quick actions from the toolbar dropdown: * **Open in Drive** -- Jump directly to the connected Drive folder * **Disconnect** -- Remove the Drive connection When exporting across batch runs, FLORA automatically deduplicates filenames to prevent overwriting previous exports in the same Drive folder. *** ## Importing from Google Drive FLORA also supports importing files from Google Drive directly onto your canvas. ### How to Import {% stepper %} {% step %} **Open the Integrations Panel** Access the Google Drive tab from the integrations panel in the sidebar. {% endstep %} {% step %} **Connect Your Account** If you haven't already, connect your Google account via OAuth. The same account used for export can be reused for import. {% endstep %} {% step %} **Browse and Select Files** Click **Import** to open the Google Drive file picker. Browse or search your Drive, then select the files you want to bring onto your canvas. {% endstep %} {% step %} **Files Appear on Canvas** Selected files are imported and placed onto your canvas as source nodes, ready to be connected to generation nodes, techniques, or other workflow steps. {% endstep %} {% endstepper %} *** ## Importing from Google Drive In addition to exporting, FLORA supports importing files directly from Google Drive into your workspace. ### How to Import 1. Open the **Library** panel from the toolbar 2. Select the **Google Drive** tab 3. Browse or search your Drive for files to import 4. Select the files you want and click **Import** Imported files go through FLORA's standard upload pipeline and appear as assets in your workspace, ready to use on the canvas. ### Managing Imported Files Hover over any imported Drive asset to access management actions: * **Rename** — Change the asset's display name in FLORA * **Download** — Save a local copy of the file * **Delete** — Remove the imported asset from your workspace ### Background Processing Drive imports continue running in the background even after you close the Library panel. A toast notification confirms when each batch of imports is complete, so you can keep working on the canvas while files are being imported. *** ## How It Fits Into Workflows The Export Node is a **terminal node** -- it has inputs but no outputs. It sits at the end of your workflow chain. ### Basic Workflow ``` [Text Prompt] --> [Image Node] --> [Export Node: Google Drive] ``` Generate an image and automatically push it to your Drive. ### Multi-Step Workflow ``` [Image Upload] --> [Image Node: Upscale] --> [Color Grade Action] --> [Export Node] ``` Upscale, color-grade, and export in one connected pipeline. *** ## Tips * **Name your Export Nodes** -- Click the label at the top of the node to rename it. This helps when you have multiple export destinations in one project. * **Test with one file first** -- Verify your Drive connection and folder selection work before running a large batch export. * **Use the context menu** -- Right-click any node with output and select **Export to Drive** for a quick one-off export without placing an Export Node. *** ## Limitations | Limitation | Details | | ---------------------------- | ----------------------------------------------------------------------------------------------------------------------------- | | External destinations | Google Drive (import and export) and [Shopify](/more/export-to-shopify) (export only) are the supported external destinations | | Terminal node | The Export Node has no outputs -- it cannot feed into other nodes | | Not available in Techniques | Export Nodes cannot be used inside Technique workflows | | Requires integration feature | Google Drive export requires the integrations feature to be enabled for your workspace | # Router Node Route one set of inputs out to many nodes with a single connection The Router Node lets you collect a set of source nodes once and fan them out to as many downstream nodes as you like. Instead of wiring the same images, videos, or text into every node that needs them, connect those sources into a Router and point your downstream nodes at the Router's single output. Update the sources in one place and every consumer stays in sync. ### Overview The Router Node is built for **one-to-many** workflows where the same inputs feed several different operations: * **Reuse a reference set** across multiple generation nodes without reconnecting each one * **Feed shared inputs** into both an image model and a text/LLM node for analysis * **Tidy up a busy canvas** by collapsing a tangle of edges into a single hub * **Swap inputs in one place** and have the change flow to every downstream node at once Think of it as a junction: many connections in, one connection out, and that one output can be branched to wherever you need it. {% hint style="info" %} The Router is a pass-through. It doesn't generate or transform anything—downstream nodes see the original source nodes exactly as if they were connected directly. {% endhint %} *** ### Getting Started {% stepper %} {% step %} **Add the Node** From the node menu, select **Router** to add it to your canvas. {% endstep %} {% step %} **Connect your sources** Drag from your source nodes (image, video, text, audio, or document) into the Router's input handle on the left. The Router locks to the modality of what you connect. {% endstep %} {% step %} **Branch the output** Drag from the Router's output handle on the right to each node that should receive those inputs. One Router output can connect to as many downstream nodes as you need. {% endstep %} {% step %} **Run downstream** Run any downstream node as usual. It processes the routed sources exactly as if they were wired in directly. {% endstep %} {% endstepper %} *** ### How Routing Works Routers are **transparent**. When a downstream node looks at its inputs, it resolves *through* the Router to the actual source nodes text blocks, image blocks, collections, and so on. A chain like: ``` [Source] → [Router] → [Image Node] ``` behaves identically to connecting the source straight to the Image Node. The Router adds no processing, no cost, and no extra step to your generation it only manages how the connection is drawn. This is what makes one-to-many reuse possible: a single Router output can branch to many nodes, and each of those nodes still receives the genuine upstream sources. *** ### The Router Interface The Router Node is intentionally compact: #### Label Every Router shows an editable label (default: **Router**). Double-click to rename it—useful when you have several Routers and want to track what each one carries (e.g. "Character refs", "Brand palette"). #### Modality Below the label, the Router displays its current modality: * **None** — no sources connected yet * **Text**, **Image**, **Video**, **Audio**, or **Document** — once a source is connected For image and video sources, the Router shows a thumbnail preview of the first item so you can identify it at a glance. #### Handles * **Input handle (left)** — accepts your source connections * **Output handle (right)** — branches out to downstream nodes The output handle reflects its state: an empty circle when nothing is connected, a **+** once the Router has input and is ready to branch, and a filled dot when it's connected downstream. *** ### Modality A Router carries a single modality at a time, matching the sources you connect: * Connect image sources → the Router becomes an **image router** * Connect video sources → a **video router** * Connect text sources → a **text router** {% hint style="info" %} A Router stays **None** until it has a source connected, and it can't connect to a downstream node until its modality is settled. Connect your sources first, then branch the output. {% endhint %} *** ### Connecting Routers #### What a Router can connect to A Router's output can feed most generation and processing nodes Image, Video, Text, and Audio nodes just like a direct connection would. Elements can also be connected into a Router. #### Chaining Routers A Router's output can feed another Router. Chained Routers stay transparent end-to-end, so the final downstream node still resolves all the way back to the original sources. #### Current limitations | Limitation | Details | | -------------------------- | ------------------------------------------------------------------------------------------------------------------------ | | No Batch connection | A Router can't connect into a Batch Node yet. Connect sources directly to the Batch, or wire the Router to blocks first. | | No Layer Editor connection | Routers can't connect to Layer Editor nodes—connect the source directly. | | Single modality | A Router carries one modality at a time, matching its connected sources. | | Settles before branching | A Router must have a source connected (modality settled) before it can connect downstream. | *** ### Workflow Examples #### Example 1: Reuse a reference set across nodes ``` [Char sheet] ─┐ [Outfit ref] ─┼─→ [Router] ─┬─→ [Nano Banana] "generate scene" [Moodboard] ──┘ (3 images) ├─→ [Text Node] "describe the references" └─→ [Flux Kontext] "restyle" ``` Connect the reference set into a Router once, then branch it to every node that needs it. No reconnecting the same images three times. #### Example 2: Tidy a busy canvas ``` [A] ─┐ [B] ─┼─→ [Router] ─→ [Image Node] [C] ─┤ [D] ─┘ ``` Collapse many parallel edges into a single, clean connection into your generation node. #### Example 3: Update sources in one place ``` [Brand palette v2] ─→ [Router] ─┬─→ [Node 1] ├─→ [Node 2] └─→ [Node 3] ``` Swap what's feeding the Router and every downstream node picks up the change—no need to touch each connection. *** ### Tips for Effective Routing * **Name your Routers** — a clear label ("Character refs", "Style set") makes a multi-Router canvas easy to read * **Connect sources first** — the Router needs a settled modality before it can branch downstream * **Branch freely** — one Router output can feed as many nodes as you need * **Use it to declutter** — when several nodes share the same inputs, a Router turns a web of edges into one tidy hub *** ### Common Use Cases * **Shared reference libraries** — feed the same character, outfit, and environment references into every generation node in a workflow * **Parallel operations on one input set** — run the same sources through several models or prompts at once * **Analysis alongside generation** — point both an image model and a text/LLM node at the same references * **Canvas cleanup** — consolidate dense connection clusters into a single readable junction # Techniques Pre-built AI workflows that combine multiple generation steps into reusable templates Techniques are pre-built, multi-step AI workflows that you can add to your canvas and run with a single click. Each technique encapsulates a complex generation pipeline—combining multiple models and processing steps—into a simple node with defined inputs and outputs. {% embed url="" %} ## Overview Instead of manually building complex workflows from scratch, Techniques let you leverage proven creative pipelines built by FLORA and the community. Think of them as "recipes" for AI generation—you provide the ingredients (inputs), and the technique handles all the intermediate steps to produce the final result. **What makes Techniques powerful:** * **Multi-step workflows** - Chain multiple AI models together (e.g., analyze image → generate prompt → create video) * **Action nodes** - Include deterministic processing steps (color grading, audio extraction, video effects) alongside AI generation * **Batch processing** - Include Batch nodes inside techniques to fan out and process collections of items in a single run * **Consistent results** - Same pipeline, same quality, every time * **Time-saving** - Skip the setup and get straight to creating * **Curated quality** - Each technique is tested and optimized for its specific use case *** ## The Techniques Library ### Browsing Techniques Access the Techniques Library from the sidebar to explore all available techniques. The library features: * **Featured carousel** - Highlighted techniques at the top * **Category filters** - Browse by use case * **Search** - Find techniques by name, description, or tags * **Preview cards** - See sample outputs before adding to canvas ### Categories Techniques are organized into the following categories: | Category | Description | | --------------------------- | ---------------------------------------------- | | Brand & Visual Design | Logo treatments, brand assets, visual identity | | Product Visualization | Product shots, mockups, lifestyle imagery | | Marketing & Ads | Ad creatives, social media content, campaigns | | Video & Animation | Motion graphics, animated sequences, video ads | | Fashion & Apparel Editorial | Lookbooks, fashion photography, styling | | Content Packaging | Thumbnails, covers, promotional materials | | Print Film/VFX | Film-quality effects, cinematic treatments | | Space & Architecture | Interior design, architectural visualization | | Fun & Inspiration | Creative experiments, artistic styles | ### Technique Details Click any technique card to view its detail page, which shows: * **Description** - What the technique does and ideal use cases * **Inputs required** - What you need to provide (images, videos, or text) * **Outputs produced** - What you'll receive when the technique completes * **Example presets** - Sample inputs and their corresponding outputs * **Usage cost** - How much usage the technique consumes per run * **Estimated time** - Approximate execution duration *** ## Adding Techniques to Your Canvas ### From the Add-Node Menu The fastest way to add a technique is directly from the canvas add-node menu: 1. Open the add-node menu (click **+** on the toolbar or press **Shift+T** to jump straight to techniques) 2. Select **Technique** from the Media section to open the technique picker panel 3. Browse **Recent** and **Featured** techniques, or use the search bar to find a specific one 4. Hover over any technique to see a preview card with its thumbnail, description, and creator info 5. Click a technique or press its number key shortcut (**1-9**) to add it to your canvas at the cursor position Use the **Back** button to return to the main add-node menu, or **View all** to open the full Techniques Library. ### From the Library {% stepper %} {% step %} **Browse and Select** Open the Techniques Library and find a technique that matches your needs. Click the technique card to view details. {% endstep %} {% step %} **Add to Canvas** Click the **Add to Canvas** button. A Technique Node node appears on your canvas with all input and output slots ready. {% endstep %} {% step %} **Connect Inputs** Provide the required inputs by connecting source nodes to each input slot on the technique node. {% endstep %} {% step %} **Run the Technique** Once all inputs are connected, click **Run** to execute the technique. The workflow processes automatically and creates output nodes when complete. {% endstep %} {% endstepper %} *** ## The Technique Node When you add a technique to your canvas, it appears as a Technique Node—a specialized node that represents the entire workflow. ### Node Structure * **Header** - Displays the technique name, icon, and usage cost * **Input slots** - Connection points for each required input (left side) * **Output slots** - Connection points for each output (right side) * **Run button** - Execute the technique when all inputs are ready ### Input Cards Each input slot shows: * **Input name** - What type of content is expected * **Input type** - Image, video, or text indicator * **Connection status** - Whether an input is connected or missing * **Thumbnail** - Preview of the connected input (for images/videos) *** ## Providing Inputs Techniques require specific inputs to run. There are several ways to provide each input: ### Connect Existing Nodes Drag a connection from any compatible node output to the technique's input slot: * **Image inputs** - Connect from Image nodes, uploads, or other image outputs * **Video inputs** - Connect from Video nodes or video uploads * **Text inputs** - Connect from Text nodes * **Document inputs** - Connect from Document nodes (PDF to Text, PDF to Image) ### Quick Input Actions Right-click an input slot or use the input menu to quickly add content: | Action | Description | | ---------- | --------------------------------------------------- | | Add Preset | Insert a sample input from the technique's examples | | Generate | Run the technique and generate outputs | | Upload | Upload a file directly to this input | ### Using Presets Each technique includes example presets—sample inputs that demonstrate the technique's capabilities. Adding a preset creates a static node with the sample content, letting you test the technique immediately. *** ## Running Techniques ### Execution Process When you run a technique: 1. **Validation** - FLORA checks all required inputs are connected and valid 2. **Usage check** - Confirms you have sufficient budget for the run 3. **Processing** - The internal workflow executes step by step (including both AI generation steps and any deterministic action steps) 4. **Progress tracking** - The technique node shows real-time progress 5. **Output creation** - Result nodes appear connected to the technique's outputs ### Parameter Overrides Some techniques allow you to customize specific outputs before running. When a technique author has enabled **user-editable parameters** on an output, you'll see a settings icon on that output slot. Click it to override the model or generation parameters (such as aspect ratio, quality, or style) for that output. Overrides are validated server-side — only safe parameters can be changed (e.g., you cannot override cost-related knobs like quantity or seed). The technique's usage cost is automatically recalculated to reflect your chosen overrides. ### Usage Costs Each technique has a usage cost displayed in the header. This cost covers all the internal generation steps—you pay once for the entire workflow, not per internal step. If you override parameters on any output, the cost is recalculated to reflect the actual models and settings used. {% hint style="info" %} Usage is deducted when the technique starts. If a technique fails, the deduction is automatically refunded. {% endhint %} ### Progress Indicator While running, the technique node displays: * **Progress bar** - Visual indicator of completion percentage * **Current step** - Which internal node is currently processing * **Estimated time remaining** - Approximate time until completion *** ## Working with Outputs ### Output Nodes When a technique completes, it automatically creates output nodes on your canvas: * **Image outputs** - Appear as result image nodes * **Video outputs** - Appear as result video nodes * **Text outputs** - Appear as result text nodes These output nodes are connected to the technique node and can be used as inputs for other nodes in your workflow. ### Collection Outputs Techniques that include Batch nodes produce **collection outputs** — multiple items from a single output slot. When a technique contains a fan-out step, the batch results are surfaced as a scrollable collection on the technique node's output. Each item in the collection can be viewed individually in the sidebar, detail view, or App Mode. ### Feedback After a technique completes, you can provide feedback on the outputs using the thumbs up/down buttons. This helps improve technique quality over time. ### Chaining Techniques Output nodes from one technique can be connected to inputs of another technique or any other node, enabling complex multi-technique workflows. *** ## Batch Processing with Techniques Techniques can be run through a [Batch Node](/nodes/batch-node) to process multiple inputs through the same technique pipeline at once. This is useful when you want to apply the same creative workflow to a collection of images, videos, or other media. ### How It Works 1. Add a **Batch Node** to your canvas and populate it with your items 2. Connect the Batch Node's output to a **Technique Node's** input 3. Click **Run** on the technique — each batch item is processed through the full technique pipeline independently 4. Outputs are created for each batch item, with carousel navigation to browse results ### Use Cases * **Product line processing** - Run a lifestyle photography technique across an entire product catalog * **Campaign variations** - Generate branded ad creatives from multiple source images in one pass * **Bulk video creation** - Convert a batch of product shots into animated videos using the same technique *** ## Detaching Techniques Want to see how a technique works internally? You can "detach" a technique to expand it into its constituent nodes. ### How to Detach 1. Select the Technique Node 2. Click the **Detach** option in the toolbar or context menu 3. The technique expands into individual nodes on your canvas ### What Happens * The technique node is replaced by all internal nodes * Connections are preserved and mapped to the expanded nodes * You can now see and modify each step of the workflow * Action nodes appear as labeled "Action" blocks — the label is derived from the action's function name or schema {% hint style="warning" %} Detaching is one-way—you cannot re-collapse nodes back into a technique node. Consider duplicating the technique before detaching if you want to keep the original. {% endhint %} *** ## Viewing a Technique's Workflow You can preview a technique's internal workflow without detaching it by clicking **View workflow** in the Technique Node toolbar. This opens a read-only overlay that shows the full graph — generation nodes, action nodes, and their connections. Action nodes are displayed with a distinctive violet style so they're easy to spot. *** ## Tips for Using Techniques * **Check the examples first** - Review technique presets to understand what inputs work best * **Match input quality** - Higher quality inputs generally produce better outputs * **Mind aspect ratios** - Some techniques work best with specific aspect ratios (noted in input descriptions) * **Test with presets** - Try the built-in presets before using your own content * **Watch usage costs** - Complex techniques with many steps may use more of your budget * **Use technique outputs downstream** - Chain technique outputs into other nodes for extended workflows *** ## Technique vs. Manual Workflows | Aspect | Technique | Manual Workflow | | -------------- | ----------------------------------- | --------------------------------- | | Setup time | Instant—just add and connect inputs | Requires building from scratch | | Consistency | Same pipeline every time | May vary between sessions | | Customization | Author-defined parameter overrides | Full control over every parameter | | Learning curve | Use immediately without expertise | Requires understanding each model | | Usage cost | Single fixed cost | Varies based on nodes used | *** ## Common Use Cases * **Product photography** - Transform product shots into lifestyle imagery * **Ad creation** - Generate multiple ad variations from a single product image * **Video production** - Convert static images into animated video content * **Brand assets** - Create consistent branded materials from reference images * **Social content** - Produce platform-optimized content from source materials * **Concept exploration** - Quickly test creative directions without manual setup *** ## App Mode Techniques can also be run outside of the canvas using App Mode. This provides a streamlined, focused interface where you simply provide inputs and run the technique—no canvas setup required. App Mode is ideal for quick, repeated use of your favorite techniques without the overhead of managing a full project. *** ## Technique Builder Technique Builder lets you package any multi-step workflow on your canvas into a reusable Technique — define inputs, define outputs, and publish. You can also edit published Techniques to update their workflow, inputs, outputs, or listing details. {% hint style="info" %} For the full guide on creating, editing, and publishing your own Techniques, see [**Technique Builder**](/nodes/technique-builder). {% endhint %} # Technique Builder Turn any canvas workflow into a reusable, shareable Technique Technique Builder lets you package any multi-step workflow on your canvas into a reusable Technique. Define what goes in, what comes out, and publish it — so you or anyone else can reuse it as a single node on canvas or as a standalone app. {% embed url="" %} *** ## Overview If you've ever built a workflow you want to reuse — or share with your team — Technique Builder is how you turn it into a first-class Technique. **What you can do:** * **Compress workflows** — Collapse a multi-node pipeline into a single reusable step * **Publish to your Workspace** — Make it available in the Techniques Library for your team * **Reuse anywhere** — Drop your published Technique onto any canvas like a regular node * **Share as an app** — Others can run your Technique through a simplified app interface, no canvas required *** ## Getting Started To open Technique Builder, click the **Build Technique** button at the bottom of the canvas. This button appears when you have a workflow on your canvas that's ready to be packaged. {% hint style="info" %} When you enter Technique Builder mode, your canvas nodes become **locked for editing**. This is by design — the builder needs a stable snapshot of your workflow to define inputs and outputs. You can exit at any time to resume editing. {% endhint %} {% hint style="info" %} In **create mode**, the left tool palette (Add Node, Library, Techniques, History, etc.) is hidden to keep the focus on defining your technique's inputs and outputs. The palette returns when you exit the builder or switch to edit mode. {% endhint %} *** ## The Builder Flow Technique Builder walks you through four steps: | Step | What you do | | ----------- | ----------------------------------------------------- | | **Intro** | Learn what Technique Builder does and how it works | | **Input** | Select which nodes serve as inputs to your Technique | | **Output** | Select which nodes serve as outputs of your Technique | | **Publish** | Fill in details and publish your Technique | A progress bar at the top of the panel tracks your current step. You can navigate back to previous steps at any time. *** ## Step 1: Intro The intro screen gives you a quick walkthrough of what Technique Builder enables: 1. **Compress your creative workflow into a single step** 2. **Publish it to the Workspace library** 3. **Reuse it anywhere on canvas** 4. **Share with others as a simple app** Click **Start Building** to begin defining your Technique's inputs and outputs. *** ## Step 2: Define Inputs **"What should be the inputs?"** Select the nodes on your canvas that should serve as the inputs to your Technique. Inputs are what the user provides when they run the Technique — for example, a reference image, a text prompt, or a style image. ### How input selection works When you're on the Input step, **click nodes directly on the canvas** to select them as inputs. A node qualifies as an input candidate if: * It has **no incoming connections** (i.e., it's a starting node in your workflow) * It has **generated output content** (an image, video, or text result) * It's a **supported node type** (standard generation nodes — not groups, comments, layer editors, collections, or technique nodes) Selected inputs appear as cards in the builder panel. You must select **at least one input** to continue. ### Configuring each input For each selected input, you can customize: * **Name** — Give the input a clear, descriptive label (e.g., "Reference Photo", "Style Prompt"). This is what users will see when running your Technique. Names are required and must not be empty. * **Description** (optional) — A short explanation of what this input expects (max 40 characters). This helps users understand what to provide. * **Preset** — A preview of the current content on this node. When someone uses your Technique, this content serves as the default/example value. To remove an input, click the trash icon on its card. *** ## Step 3: Define Outputs **"What should be the outputs?"** Select the nodes on your canvas that represent the final results of your workflow. Outputs are what the Technique produces — the generated images, videos, or text that users will receive. ### How output selection works Click nodes on the canvas to select them as outputs. A node qualifies as an output candidate if: * It has **no outgoing connections** (i.e., it's an end node in your workflow) * It has **generated output content** * It's a **supported node type** You must select **at least one output** to continue. ### Configuring each output Output cards work the same as input cards: * **Name** — A descriptive label for the output (e.g., "Final Render", "Upscaled Video") * **Description** (optional) — What this output produces (max 40 characters) * **Preset** — A preview thumbnail of the current output content * **User-editable parameters** (toggle) — When enabled, viewers running your Technique can override the model and generation parameters on this output at run time. This is useful when you want to give users control over quality, style, or model choice for specific outputs while keeping the rest of the workflow fixed. The toggle is only available on outputs backed by generation nodes that have configurable parameters. ### Including Batch Nodes Techniques support Batch nodes for fan-out processing. When your workflow includes a Batch node, the builder records it as part of the technique graph. At runtime, the batch fan-out executes inside the technique and the resulting collection outputs are surfaced to the user. {% hint style="warning" %} Explosive batch configurations — such as chaining multiple batch nodes that would multiply into an excessive number of tasks — are blocked at publish time. Keep batch workflows linear to avoid validation errors. {% endhint %} ### Graph validation When you click **Continue** from the outputs step, the builder validates your Technique's graph to ensure it's well-formed. Validation checks include: * All inputs and outputs are properly connected * No circular dependencies exist in the workflow * All referenced nodes have valid configurations * Node types are compatible with the Technique format * Batch configurations do not create explosive fan-out chains If validation fails, you'll see an error message explaining what needs to be fixed before you can proceed. *** ## Step 4: Publish **"Publish your Technique"** Review your Technique and fill in the details that help others discover, understand, and use it. ### Name The name of your Technique as it will appear in the library and on canvas. ### Description A description of what this Technique generates and how to use it (max 200 characters). This appears as a subtitle in the library. Name and description are both required to publish. ### Usage Cost The total usage cost to run your Technique, automatically calculated from the models and generation steps in your workflow. This is displayed to users before they run the Technique. ### Thumbnail A preview image for your Technique in the library. The builder automatically generates a thumbnail from your output nodes, but you can upload a custom one by clicking **Upload**. ### Category (optional) Assign your Technique to a category so users can find it when browsing the library: | Category | Best for | | --------------------------- | ------------------------------------------------------ | | Brand & Visual Design | Logo, brand identity, and visual design workflows | | Product Visualization | Product shots, mockups, and 3D renders | | Marketing & Ads | Ad creatives, campaign assets, and promotional content | | Video & Animation | Video generation, animation, and motion graphics | | Fashion & Apparel Editorial | Fashion shoots, lookbooks, and apparel visualization | | Content Packaging | Social media, thumbnails, and content formatting | | Film & VFX | Film production, VFX, and cinematic content | | Space & Architecture | Interior design, architecture, and spatial design | | Fun & Inspiration | Creative experiments and artistic exploration | ### Tags (optional) Add keyword tags to improve discoverability (e.g., `marketing`, `ads`, `film`). Type a tag and press **Enter**, **Tab**, or use a comma to add it. You can also paste multiple comma-separated tags at once. Remove tags by clicking the **x** on each badge. ### App Link A preview of the shareable URL for your Technique: `flora.ai/technique/your-technique-name`. The slug is automatically generated from the Technique name. ### Visibility Control who can access your Technique: | Visibility | Access | | ------------- | ------------------------------------------------------------------------------------------------------------- | | **Private** | Only you can access. This is the default. | | **Workspace** | Everyone in your workspace can access. | | **Community** | Listed in the Community Library for anyone to discover and use. Reviewed by the FLORA team before publishing. | {% hint style="info" %} **Community review:** When you select Community visibility, your Technique is submitted for review by the FLORA team before it becomes publicly listed. You'll be notified once it's approved. {% endhint %} *** ## Publish Success After publishing, a confirmation dialog appears showing your Technique's thumbnail and status. The dialog varies based on your chosen visibility: * **Private** — "Your Technique is now in your Private Library." You can change the app link access from here. * **Workspace** — "Your Technique is now in your Workspace Library." You can adjust who can access the app link. * **Public (submitted for review)** — "Your Technique is now in review." A shareable URL is shown with a copy button. From the publish success dialog you can: * **Copy the app link** to share your Technique with others * **Change app link access** — control whether the app link is accessible to only you, workspace members, or everyone (for private and workspace techniques) * **Open the app** — launch the Technique's app-mode interface in a new tab * **Return to canvas** — close the dialog and go back to editing *** ## Editing a Published Technique After publishing, you can edit your Technique to update its workflow, inputs, outputs, or listing details. Editing updates the existing Technique in place — it doesn't create a separate copy. ### How to start editing Hover over a Technique node on your canvas to reveal its toolbar. If you're the Technique's creator, you'll see an **Edit technique** button (pencil icon). Click it to enter edit mode. When you enter edit mode, the Technique node is **detached** — its internal workflow is expanded back onto your canvas as individual nodes inside a group, so you can modify them directly. A snapshot of the original state is saved so your changes can be reverted if you cancel. ### The edit flow Editing walks you through four steps, shown in a progress bar at the top of the panel: | Step | What you do | | ------------ | ------------------------------------------------------------------------- | | **Workflow** | Edit prompts, update nodes, connect or disconnect nodes | | **Input** | Review and update which nodes serve as inputs | | **Output** | Review and update which nodes serve as outputs | | **Publish** | Update listing details (name, description, thumbnail, etc.) and republish | #### Step 1: Edit Workflow Your Technique's workflow is expanded onto the canvas. The canvas is unlocked during this step — you can: * Edit prompts and model parameters on any node * Change models on generation nodes * Add or remove nodes from the workflow * Connect or disconnect nodes to restructure the pipeline Click **Continue** when you're done editing the workflow to proceed to input selection. #### Steps 2–4: Input, Output, Publish These steps work the same as when creating a new Technique. The builder pre-populates your previous input/output selections and listing details, so you only need to adjust what's changed. When you click **Publish**, the existing Technique is updated. Anyone using it will get the new version the next time they run it. ### Cancelling an edit Click **Cancel** or the **X** button at any time to discard your changes. The canvas reverts to its original state — the detached nodes are removed and the original Technique node is restored with its previous connections. ### Toolbar actions The Technique node toolbar also provides: * **Detach into nodes** — Expands the Technique into individual canvas nodes without entering edit mode. Useful for forking a Technique into a standalone workflow. * **View workflow** — Opens a read-only overlay showing the Technique's internal workflow graph. Action nodes appear as violet "Action" blocks so they're easy to distinguish from generation steps. *** ## Tips & Best Practices * **Run your workflow first** — Make sure all nodes have generated outputs before entering Technique Builder. Nodes without output content can't be selected as inputs or outputs. * **Use clear names** — Input and output names are what users see when running your Technique. Be specific: "Product Photo" is better than "Input 1". * **Keep descriptions short** — You have 40 characters. Focus on what the user needs to know: "Upload a front-facing product shot" tells them exactly what to provide. * **Choose meaningful outputs** — Select the final, polished results — not intermediate steps. Users expect the output to be the finished product. * **Test before publishing** — Ensure your workflow produces consistent, high-quality results before packaging it as a Technique. * **Pick the right category** — Correct categorization helps users find your Technique when browsing the library. *** ## Supported & Unsupported Node Types Techniques only support a specific set of node types. Your entire workflow — inputs, outputs, and all intermediate nodes — must be built using supported nodes. **Supported node types:** * Image nodes (text-to-image, image-to-image) * Video nodes (text-to-video, image-to-video) * Text nodes (text-to-text, image-to-text, video-to-text) * Action nodes (code blocks) * Element nodes * Static image blocks * Empty image blocks * Document nodes (PDF to Text, PDF to Image) * Router nodes (routers are included transparently — they function within the technique but are not exposed as user-facing inputs or outputs) * Batch nodes (fan-out processing within a technique) * Static image nodes * Empty image nodes {% hint style="info" %} **Action nodes in techniques** execute their code via the sandbox during the technique run. Their outputs flow through to downstream nodes or to technique output slots via result nodes. This lets you include deterministic post-processing steps (color grading, frame extraction, audio splitting, etc.) directly inside a technique pipeline. {% endhint %} {% hint style="danger" %} **Batch nodes and Layer Editor nodes are not supported anywhere in a Technique.** If your workflow uses any of these, you'll need to rebuild it using only the supported node types listed above before it can be packaged as a Technique. {% endhint %} **Other unsupported node types:** * Comments * Groups * Inpaint/Outpaint image nodes * Result nodes (image, text, video) * Technique nodes (nested techniques) {% hint style="warning" %} **Intermediate nodes are preserved.** Nodes that sit between your selected inputs and outputs are automatically included in the Technique graph — you don't need to select them. The builder traces the connections from outputs back to inputs and retains all nodes along the path. {% endhint %} *** ## Exiting the Builder You can exit Technique Builder at any time by clicking the **X** button in the panel header or clicking **Go Back** on the intro screen. If you have unsaved changes (selected inputs/outputs or publish details), you'll be asked to confirm before discarding your draft. # Editor ## Editor The Editor is the core of Flora. It is a powerful canvas with a toolbar where your creative process happens. The Editor consists of a few main areas: {% stepper %} {% step %} [**Canvas**](/editor/canvas) {% endstep %} {% step %} [**Toolbar**](/editor/toolbar) {% endstep %} {% step %} [**Navigation**](/editor/navigation) {% endstep %} {% step %} [**Styles (deprecated)**](/faqs-how-to.../styles) {% endstep %} {% endstepper %} # Canvas ## Summary FLORA's canvas is your creative garden. You can spawn nodes, connect nodes, edit parameters, and create your own personalized workflows.
## Getting Started There are many ways to start populating the canvas with your creative thoughts. Simply **Double-Click** to spawn a node or click on any of the selected workflows to start.
## Connecting Nodes FLORA brings over 50+ text, image, and video models to one canvas, saving you time and effort in context- and tool-switching. With this approach, you can connect nodes of different modalities, using our open-ended canvas to ideate, explore different creative directions, compare models and prompts, and more. To connect a node, simply drag the '+' handle from a node that's on the canvas. Here are **just a couple** of example use cases of connecting nodes: * **Text node to image node:** Connect the output of a text node to the input of an image node. * **Image node to video node:** Connect one or multiple image nodes to a video node as a reference frame (depending on the model, can be first frame, last frame, or both). As AI model capabilities advance, the number of possibilities around combining text / image / video nodes expands. ## Color Tagging You can assign color tags to nodes to visually organize your canvas. Color tags appear as colored highlights on nodes, making it easy to group related content by theme, status, or any system that works for your workflow. ### Tagging individual nodes Select a node and use the **color tag** option in the node toolbar to assign a color. ### Tagging multiple nodes at once Select multiple nodes on your canvas (drag-select or Shift+click), and the **multi-selection toolbar** will appear with a color tag picker. Choose a color to apply it to all selected nodes simultaneously — useful for quickly categorizing batches of nodes. ## Bulk Parameter Editing When you select **two or more blocks of the same type and mode** on the canvas (e.g., two image generation blocks both using the same modality), a **Bulk Parameters Panel** appears in the right sidebar. This lets you edit shared parameters across all selected blocks at once — no need to update each block individually. * **Shared parameters only** — The panel shows only the parameters that exist on every selected block. Parameters unique to a single block are not shown. * **Mixed values** — When selected blocks have different values for a parameter, the field shows a "Mixed" indicator. Editing the field applies the new value to all selected blocks. * **Model selector** — If the selected blocks share compatible models, you can switch models for all of them at once from the bulk panel. This is especially useful when you want to apply consistent settings (seed, guidance, aspect ratio, etc.) across multiple blocks in a comparison or batch workflow. ### Reordering Connected Inputs When a node has multiple connected inputs — such as several images wired into a generation node, or multiple text inputs on an Action node — you can **drag to reorder** them. Grab any connected input and drag it up or down to change the order in which inputs are processed. ### Wiring In-Progress Nodes You can connect nodes that are still processing (such as a Document node that hasn't finished extracting text yet) to downstream nodes. FLORA validates readiness at generation time, not at connection time — so you can build your workflow while earlier steps are still running. # Toolbar ## Summary The toolbar is where you can find the main editor tools.
## Spawning Nodes By clicking the + button on the toolbar you can spawn different types of nodes: Text, Image, Video, and Audio. The add-node menu also includes a **Technique** entry — select it or press **Shift+T** to open a technique picker panel where you can browse, search, and add techniques directly to your canvas.
## Assets By clicking on the Assets button, you can browse all the visual assets you uploaded to the canvas.
## Generation History The Generation History button stores all the generated assets from your account, making it seamless for you to transport assets between projects.
## Flows Flows allow you to explore our pre-made workflows for various creative use cases. By **hovering** over the Flows button, you can view recent workflows along with the number of nodes deployed. **Clicking** on the Flows button lets you browse curated categories to find workflows that inspire your creative process.
## Split Into Layers The **Split Into Layers** tool lets you decompose an image into separate layer components. Select an image node and click the Split Into Layers button in the toolbar to open a visual cascade picker. Choose how many layers to split into (2–5) from the overlay, and FLORA automatically separates the image into distinct layers that you can edit independently. ## Comment The Comment Button is where you activate the comment cursor. Click on anywhere to start a text bubble and leave a comment for your collaborators.
# Image Editing ## Overview FLORA provides powerful image editing capabilities that allow you to modify and extend your generated images directly on the canvas. The Image Editor includes two main features: * **Inpainting**: Paint over areas of an image to regenerate specific regions * **Outpainting**: Expand the canvas beyond the original image boundaries These tools work seamlessly with FLORA's AI generation pipeline, letting you iteratively refine your images without leaving the editor. {% embed url="" %} *** ## Inpainting Inpainting allows you to selectively regenerate portions of an image by painting a mask over the areas you want to change. This is useful for: * Removing unwanted objects from an image * Replacing specific elements with something new * Fixing imperfections or artifacts * Adding new objects to a scene ### How to Use Inpainting {% stepper %} {% step %} **Select an Image Node** Click on an image node that contains a generated or uploaded image. {% endstep %} {% step %} **Activate the Inpaint Tool** Click the **Inpaint** tool in the image editing toolbar. Your cursor will change to a brush indicator. {% endstep %} {% step %} **Paint the Mask** Click and drag on the image to paint over the areas you want to regenerate. The painted areas will be highlighted with a green crosshatch pattern. **Brush Controls:** * **Brush Size**: Adjust the brush size using the slider in the brush panel (5-100 pixels) * **Undo Strokes**: Press `Ctrl/Cmd + Z` to undo individual brush strokes * **Clear All**: Remove all painted strokes to start over {% endstep %} {% step %} **Enter Your Prompt** In the prompt input that appears below the image, describe what you want to generate in the painted area. Example prompts: * "A red sports car" (to replace a vehicle) * "Blue sky with clouds" (to fix the background) * "Remove the object" (to seamlessly blend the area) {% endstep %} {% step %} **Generate** Click the submit button to generate. A new image node will be created with your inpainted result, preserving your original image. {% endstep %} {% endstepper %} ### How Inpainting Works When you submit an inpainting request: 1. **Mask Generation**: Your brush strokes are converted into a mask image that tells the AI which areas to regenerate 2. **Composite Creation**: The original image and mask are combined and sent to the selected AI model 3. **Generation**: The model generates new content only within the masked regions while preserving the rest of the image 4. **Result**: A new image node appears with the inpainted result {% hint style="info" %} Different AI models may require different mask formats. FLORA automatically handles this conversion based on the model you've selected. {% endhint %} ### Tips for Better Inpainting Results * **Paint generously**: Slightly extend your mask beyond the edges of what you want to change for smoother blending * **Be descriptive**: Provide clear, detailed prompts about what should appear in the masked area * **Consider context**: The AI uses the surrounding image as context, so your prompt should fit naturally with the scene * **Iterate**: If the first result isn't perfect, try adjusting your mask or prompt and generate again ### Supported Models for Inpainting Several AI models in FLORA support inpainting: * **GPT Image 1.5** - Supports advanced mask-based inpainting * **Nano Banana Pro** - Uses outline-based inpainting * **Ideogram** - Creative inpainting capabilities {% hint style="info" %} Model availability may vary. Check the model selector for currently available options. {% endhint %} *** ## Outpainting Outpainting (also known as canvas expansion) allows you to extend an image beyond its original boundaries. The AI will generate new content that seamlessly continues from the edges of your original image. ### How to Use Outpainting {% stepper %} {% step %} **Select an Image Node** Click on an image node containing the image you want to expand. {% endstep %} {% step %} **Activate the Outpaint Tool** Click the **Outpaint** tool in the image editing toolbar. Green resize handles will appear around the image. {% endstep %} {% step %} **Expand the Canvas** Drag the handles to expand the canvas in any direction: * **Edge handles**: Drag the top, bottom, left, or right edges to expand in one direction * **Corner handles**: Drag corners for diagonal expansion {% endstep %} {% step %} **Enter Your Prompt** In the prompt input that appears, describe what should fill the expanded area. If left empty, FLORA will use "Fill this expanded area" as the default prompt. Example prompts: * "Continue the beach scene with palm trees" * "Extend the sky with sunset colors" * "Add more of the forest landscape" {% endstep %} {% step %} **Generate** Click submit to generate. A new image node will appear with your expanded image. {% endstep %} {% endstepper %} ### How Outpainting Works When you submit an outpainting request: 1. **Composite Creation**: FLORA creates a new canvas at the expanded dimensions with your original image positioned in the appropriate location 2. **Generation**: The AI model fills in the expanded regions while maintaining visual continuity with your original image 3. **Result**: A new node displays the expanded image ### Tips for Better Outpainting Results * **Expand gradually**: For very large expansions, consider doing multiple smaller expansions rather than one massive change * **Provide context**: Your prompt should describe what naturally extends from the edges of your image * **Watch the edges**: Pay attention to elements at the edges of your original image—the AI will try to continue them * **Aspect ratio**: Consider your final desired aspect ratio when expanding; you can expand more in one direction than another *** ## Workflow Tips ### Combining Editing Tools You can use inpainting and outpainting together for powerful workflows: 1. **Generate** an initial image 2. **Outpaint** to expand the canvas and add more scene 3. **Inpaint** specific areas to refine details or fix artifacts 4. **Repeat** as needed to achieve your desired result ### Preserving Your Work * Each edit creates a new image node, so your original is always preserved * You can compare results side-by-side on the canvas * Delete nodes you don't need to keep your workspace organized ### Keyboard Shortcuts | Action | Shortcut | | ----------------- | ----------------------------------------- | | Undo brush stroke | `Ctrl/Cmd + Z` | | Cancel generation | Click the cancel button during generation | # Timeline Editor Trim, sequence, caption, and render your clips into a finished video on the canvas The Timeline Editor is FLORA's video editor, built directly into the canvas. It's a real multi-track timeline for trimming, sequencing, captioning, scoring, and rendering video — so the clips you generate in FLORA go all the way to a finished MP4 without leaving your workflow. Because the editor is a **node**, the finished cut flows right back onto the canvas as a new video you can keep building on. ## Overview The Timeline Editor turns a node into a full video-editing workspace, connected live to the clips that feed it. In it you can: * **Assemble a sequence** from your generated clips, laid out end to end on a single track * **Trim and split** clips to the exact frames you want * **Layer up to 10 tracks** of video, audio, text, captions, images, and shapes * **Style each item** — position, size, crop, rotation, opacity, playback rate, volume, and fades * **Caption and title** with your choice of font, stroke, and alignment * **Render a finished MP4** server-side, saved back to the canvas as a new version * **Export to a pro editor** as an EDL / XML / FCPXML bundle with your source media This is ideal for cutting a social clip, assembling a product teaser, finishing an ad, or scoring a short — all without exporting to CapCut or Premiere just to put together what you already made in FLORA. {% hint style="info" %} The Timeline Editor reads its clips **live from the nodes wired into it** and always uses each node's latest generation. That link back to your graph is what makes it different from a standalone editor — see [Connected inputs & live updates](#connected-inputs-and-live-updates). {% endhint %} *** ## Getting Started ### Adding a Timeline Editor Node {% stepper %} {% step %} **Add the Node** From the node menu, select **Timeline** to add it to your canvas. (You can also add the node first and connect clips afterward — order doesn't matter.) {% endstep %} {% step %} **Connect Source Clips** Connect clip nodes to the Timeline node. Each connected input becomes one layer, in the order you connected them. Select several nodes to wire them all in at once. | You connect… | You get… | | ------------- | ------------------------------- | | A video node | A video clip | | An image node | A still clip (with a thumbnail) | | An audio node | An audio track | | A text node | An editable text layer | | {% endstep %} | | {% step %} **Open the Editor** **Double-click** the Timeline node, or click **Open Timeline Editor** in the node toolbar. The editor opens in place — the node becomes the editing surface, and the canvas stays pannable around it. {% endstep %} {% step %} **Edit and Render** Arrange, trim, and style your clips, then open the **Export** panel and choose **Edited File** to render a finished MP4 back onto the canvas. {% endstep %} {% endstepper %} {% hint style="info" %} The Timeline Editor is a paid feature — available on Starter and above (see [Plans & limits](#plans-and-limits)). {% endhint %} *** ## The Editor Interface The editor unfolds around the node itself — there's no full-screen takeover, and the FLORA chrome stays visible. It has three regions. ### The Preview The live playback area, shown right on the node. It reflects whatever the playhead is over. Opening reframes the canvas so the node clears the surrounding chrome; if you pan or zoom the node out of view, a **Recenter** toast brings you back. ### The Properties Sidebar On the right — a tools row plus an inspector that changes with your selection, organized into collapsible sections (at parity with the [Layer Editor](/nodes/layer-editor)). | Selected item | Controls | | ----------------------- | -------------------------------------------------------------------------------------- | | **Video clip** | Playback rate, volume, crop, rotation, rounded corners, position, size, opacity, fades | | **Audio clip** | Volume, playback rate, fade in/out, timing | | **Text / caption** | Font (Google Fonts), stroke, alignment, position, size, opacity, rotation | | **Image / shape / GIF** | Position, size, crop, rotation, opacity, rounded corners | ### The Timeline Drawer Slides up along the bottom, holding the playback controls and the timeline — tracks, clips, playhead, and ruler. Clips show **film-strip thumbnails** (and audio shows a **waveform**), and each track has **hide** and **mute** toggles to solo or silence a layer. Drag the top edge to **resize its height**, and snap its **width** between default and full canvas width. *** ## Editing Clips Your connected clips start on a **single track, end to end, in connection order** — a standard sequential assembly you then refine. ### Arranging * **Drag** a clip left or right to reorder or move it; other clips make room. * **Drag** a clip up onto another track to layer it (overlays, text, picture-in-picture). You can use up to **10 tracks**. {% hint style="info" %} **Snapping** helps clips line up against each other and the playhead. Toggle it with the `N` key or the snapping toggle. {% endhint %} ### Trimming Drag a clip's **left edge** to change where it starts, or its **right edge** to change where it ends. Trimming is non-destructive — you're changing the in/out points of that placement, not the source media. ### Splitting at the Playhead 1. Move the **playhead** to the exact frame (scrub, or click the ruler). 2. Select the clip. 3. Click the **scissors button** (**Split Clip at Playhead**). The clip splits into two, and you can trim, move, or delete either half. Split is available only when the playhead sits inside the selected clip. ### The Clip Menu Right-click a clip for its context menu: * **Go to Node** — frame the clip's source node on the canvas * **Cut** / **Copy** / **Duplicate** * **Bring to front** / **Send to back** — change stacking order where clips overlap ### Duplicating, Copying & Deleting * **Duplicate** with `Cmd/Ctrl + D`. * **Cut / Copy / Paste** with `Cmd/Ctrl + X` / `C` / `V`; **Select all** with `Cmd/Ctrl + A`. * **Delete** with `Delete` or `Backspace` (this also removes the clip's connecting edge on the canvas). ### Styling a Clip Select any item and shape it from the properties sidebar. **Every item** supports position, size, opacity, rotation, crop, layer order, and **fade in / fade out**. Video clips add playback rate, volume, and rounded corners. {% hint style="info" %} There's no keyframe animation in this version, so an item's transform values are **constant for its duration**. To change a look partway through, split the clip and style each part separately. Undo any edit with `Cmd/Ctrl + Z` — undo applies to your edits only, never the surrounding canvas. {% endhint %} *** ## Text & Captions Add titles, lower thirds, credits, and timed captions directly to your video. * **Add text in the editor** — press `T` to drop a new text item. * **Connect a text node** — it comes in as an editable text layer, linked to its source node (Go to Node works on it, and renaming the node updates the label). Style text from the sidebar: **font** (Google Fonts), **stroke** (an outline for legibility over busy footage), **alignment**, plus position, size, opacity, and layer order. Stack text above your video by putting it on a higher track. **Captions** are handled as **timestamped caption pages** — text blocks tied to points in time, so each line shows at the right moment. Style them the same way as text. *** ## Audio The Timeline Editor handles audio as its own kind of track, so you can lay music or a voiceover under your video and balance the mix. * **Connect an audio node** (including audio you generated in FLORA), **drag** an audio asset from your library, or **upload** a file in the editor. * Each source becomes its own layer, so a music bed and a voiceover can run on separate tracks. Select an audio clip to set its **volume**, **playback rate**, **fade in / fade out**, and **timing** (trim its edges and drag it to line up with the picture). Video clips carry their own audio too, adjustable per clip — lower the video or music volume to let a bed sit under dialogue, and use **Mute** during playback to check the picture on its own. *** ## Connected Inputs & Live Updates The Timeline Editor doesn't import copies of your clips — it reads them **live from the nodes wired into it**. Each connection is a layer that points back at its source node. ### Always the Latest Generation The editor reads each connected node's **current active output** — normally the newest generation. * **Regenerate while the editor is open**, and the layer refreshes to the new result automatically. * **Regenerate while it's closed**, and the layer is up to date the next time you open. * If you've selected an *older* generation as a node's active output, the editor mirrors that — it matches what the node is currently showing. {% hint style="warning" %} When a source node regenerates, its layer is **replaced, not merged** — a per-clip trim on that clip resets, because it's now different media. Trims on other clips are unaffected. {% endhint %} ### Go to Node Every clip remembers where it came from. **Right-click a clip → Go to Node**, or **double-click** it, to frame its source node. Clips are labelled by their **source node's title** (not a file name or ID), and the label updates live if you rename the node. Text clips carry this link too. ### Adding Media Without a Node * **Drag from your library** — any FLORA asset onto the timeline. Dragged-in assets are standalone (not linked to a source node). * **Upload a file** — it lands on the timeline *and* appears on the canvas as a connected media node, so it stays part of your graph. New dropped or uploaded media lands as a new track starting at 0:00 — drag it along the track to place it. ### Good to Know * **Empty inputs are allowed** — connect a node before it has generated anything; nothing appears until real content exists (you'll see a *"No generations to edit"* notice). * **Deleting a clip whose source is still connected** is temporary — it returns on reopen. Disconnect the source node to remove it for good. *** ## Playback ### The Playhead The vertical line marking the current frame; the preview always shows what it sits on. **Click** the ruler to move it, or **drag** to scrub frame by frame. While playing, the timeline follows the playhead so it stays in view. ### Playback Controls | Control | What it does | | ------------------- | -------------------------------------------------- | | Play / Pause | Plays from the playhead; pauses in place (`Space`) | | Seek bar | Drag to move through the whole composition | | Jump to start / end | Send the playhead to the beginning or the end | | Loop | Repeat playback continuously | | Mute | Silence preview audio | | Fullscreen | Expand the preview to fill the screen | Only Play / Pause has a keyboard shortcut; the rest are buttons. ### Zooming Zoom the timeline horizontally with the **zoom slider**, or by **pinching / scrolling** on a trackpad while the editor is focused. Zooming affects only the timeline, not the canvas. {% hint style="info" %} The in-editor preview is a fast approximation; the final render is produced server-side at full quality. If a clip hasn't rendered yet, the preview shows its source's first frame rather than a blank box. {% endhint %} *** ## Rendering & Export The Timeline Editor gives you two ways to get your edit out, both in the node's **Export** panel. ### Render to a Finished Video Choose **Edited File**. FLORA renders your timeline **server-side** — on FLORA's infrastructure, not in your browser — so you can keep working or close the tab while it runs. You'll see a render-time estimate up front and progress as it renders. The result is an **MP4 (H.264 video, AAC audio)**, saved back onto the canvas as a **new version of the Timeline Editor node**. {% hint style="info" %} **"Save Edits" vs. "Edited File."** *Save Edits* bakes your current timeline down to a new version. *Edited File* does the same and then downloads the MP4. Either way, a render is what produces a version. {% endhint %} ### Unexported Edits Your edits live on the node as you work, but **downstream nodes keep using the last exported version** until you export again. While you have changes that haven't been rendered, the node shows an **"Unexported edits"** badge. Downloading a cut that hasn't been rendered yet triggers a render first, then downloads the result. ### What a Render Costs Rendering is the one metered action — it draws on your workspace's **credits**, scaled by the **length and resolution** of the output. The editor shows the estimated cost before you render. Editing, previewing, and the NLE export are always free. See [How renders are billed](#how-renders-are-billed). ### Render Quality Renders use a balanced **Standard** profile by default. Where the picker is available, you can choose **High** (higher quality, slower) or **Compact** (smaller file). Set defaults per profile — output max edge, CRF, encoder preset, audio bitrate — in **Account → Timeline Editor** preferences. ### Export to a Pro Editor (NLE Bundle) To finish somewhere else, use the **Timeline + Assets** section — pick a format and download a ZIP with the project file plus your source media: | Format | Opens in | | ------------------ | ------------------------------ | | **EDL** (CMX3600) | Most NLEs | | **XML** (FCP7 XML) | Premiere Pro | | **FCPXML** | Final Cut Pro, DaVinci Resolve | The timeline is reconstructed with clips relinked to the packaged sources. This export is client-side and runs no render. {% hint style="info" %} Rendering runs server-side. If a server render isn't available and your browser falls back to in-browser encoding, use the latest **Chrome or Edge** — some browsers can't encode MP4 locally yet. {% endhint %} *** ## Plans & Limits The Timeline Editor is a paid feature. Output length and resolution scale with your plan; frame rate, format, and track/clip capacity are shared across paid plans. Rendering draws on your workspace's credits. ### Availability | Plan | Timeline Editor | | ---------- | --------------- | | Free | Not included | | Starter | ✅ | | Pro | ✅ | | Max | ✅ | | Enterprise | ✅ | If your plan doesn't include it, you'll see *"Timeline export is not included on your current plan."* ### Length, Resolution & Inputs by Plan | Limit | Starter | Pro | Max | Enterprise | | -------------- | --------------- | --------------- | ------------- | -------------- | | Max duration | 60 s | 120 s | 300 s (5 min) | 600 s (10 min) | | Max resolution | 1080p (1920 px) | 1080p (1920 px) | 4K (4096 px) | 4K (4096 px) | | Image inputs | 10 | 15 | 20 | 40 | | Video inputs | 10 | 15 | 20 | 40 | | Audio inputs | 5 | 8 | 10 | 20 | Pro adds all aspect ratios (vertical/portrait) at 1080p; Max and Enterprise unlock 4K (up to 4096×2304). ### Shared Across All Paid Plans | Limit | Value | | -------------------------------- | ---------------------------- | | Frame rate | 30 FPS | | Output format | MP4 — H.264 video, AAC audio | | Tracks (parallel layers) | 10 | | Timeline items (clip placements) | 100 | | Unique source assets | 40 | | Max video file size (per upload) | 500 MB | **Assets** are unique source files (one clip reused five times = 1 asset); **items** are placements on the timeline (that clip in five places = 5 items); **tracks** are the layers items sit on. ### How Renders Are Billed * Render cost is charged in **credits** and scales with the **length and resolution** of the output — a short 1080p cut costs a fraction of a full-length 4K one. * **Enterprise** plans include a monthly render allowance (about **$100 of render usage per user**) before overages apply. {% hint style="info" %} For the credits included with your plan and current render pricing, see the [How Pricing Works](/plans-and-billing/pricing) section. {% endhint %} ### Guardrails * **Concurrency** — you can run up to **3 renders at once per user**; additional renders queue until one finishes. * **Device playback** — preview plays in your browser, so heavy timelines can stutter on lower-memory machines. You may see a notice like *"This device supports up to 5 simultaneous video tracks — playback may stutter above that."* It's a warning, not a hard block; the final server render is unaffected. Hide or mute tracks you're not working on to smooth playback. *** ## Keyboard Shortcuts While the editor is focused, your normal canvas shortcuts are paused — so `Cmd/Ctrl + Z` undoes your last edit and never deletes the node. Shortcuts don't fire while you're typing in a text field. | Action | Shortcut | | ------------------------- | ------------------------------------------ | | Play / Pause | `Space` | | Undo | `Cmd/Ctrl + Z` | | Redo | `Cmd/Ctrl + Shift + Z` (or `Cmd/Ctrl + Y`) | | Select all | `Cmd/Ctrl + A` | | Duplicate | `Cmd/Ctrl + D` | | Cut / Copy / Paste | `Cmd/Ctrl + X` / `C` / `V` | | Delete | `Delete` or `Backspace` | | Toggle snapping | `N` | | Add text | `T` | | Add media (import) | `Cmd/Ctrl + I` | | Export EDL / XML / FCPXML | `7` / `8` / `9` (Export panel open) | **Split** has no hotkey — use the scissors button. Loop, mute, fullscreen, and jump-to-start/end are buttons, not keys. Zoom with the slider or a trackpad pinch/scroll. *** ## Limitations | Limitation | Details | | --------------------------- | -------------------------------------------------------------- | | No transitions beyond fades | No crossfades, dissolves, or wipes — only per-item fade in/out | | No effects or color | No filters, LUTs, blur, distortion, or color grading | | No keyframe animation | Per-item transforms are static for the clip's duration | | Fixed output specs | 30 FPS, MP4 / H.264 only | | Plan-capped length & size | Duration and resolution are capped by plan (60–600s, 1080p–4K) | | No Free access | The Timeline Editor is a paid feature | For anything beyond this — color, VFX, or long-form finishing — use the [NLE export](#export-to-a-pro-editor-nle-bundle) and finish in DaVinci Resolve, Final Cut, or Premiere. *** ## Troubleshooting ### A connected clip isn't showing The source node hasn't generated anything yet, or its **active generation** isn't the one you expect — the editor shows the node's currently active output. ### My trim reset itself The source node regenerated, so its layer was rebuilt from the new result. To keep a trim regardless of regeneration, drag the finished asset in from your library (standalone) instead of leaving it linked. ### A clip I deleted came back Deleting a clip whose source is still connected is temporary. **Disconnect the source node** to remove it permanently. ### Render won't start You may not be on a plan that includes it, you may be over your plan's duration/resolution cap, or you may be at the 3-renders-per-user concurrency limit. Rendering is server-side, so it continues even if you close the tab. *** ## Tips for Better Edits * **Preview freely, render once** — editing and previewing are free; you're only billed on render, so lock the cut before you export. * **Split to restyle** — since transforms are static per clip, split a clip to change a look partway through. * **Name your source nodes** — clips are labelled by node title, so clear names make a busy timeline readable. * **Hide or mute tracks** you're not working on to keep preview smooth on heavier edits. * **Use Go to Node** to jump back and regenerate a weak clip without losing your place in the edit. *** ## Common Use Cases * **Social clips** — cut generated shots into a vertical reel with captions and a music bed * **Product teasers** — assemble a 30–60s sequence, trim to the beat, and render a clean MP4 * **Ads & sizzle reels** — quick-cut multiple clips end to end with titles and audio * **Finishing hand-off** — assemble a rough cut in FLORA, then export the NLE bundle for color and VFX elsewhere # Navigation Here are some basic mouse actions to navigate in the canvas.
**Mouse and Trackpad Controls** * Left click: select, interact * Right click: menu * Right click + drag: pan the canvas * Middle mouse button: pan the canvas * Scroll wheel or trackpad: pan up, down, left and right * **\[Control^] or** **\[Command⌘]** +scroll: zoom * Pinch: zoom * Left click+drag: * On a Node: drag with mouse to move Node * On the plus button: drag connection noodle * On the canvas: box selection **Shortcuts** * Run Generation: **\[Enter]** runs selected nodes * Delete: **\[Backspace]** or **\[Delete]** while any number of nodes are selected * Copy a selected node or group of selected nodes: **\[Control^]** or **\[Command⌘]** + c * Paste a node or group of nodes, or paste any media source: image or video: **\[Control^]** or **\[Command⌘]** + v * Toggle control surfaces: **\[H]** hides or shows the prompt overlay on selected image and video nodes, useful for a cleaner view of your canvas *** ## Canvas Search Press **\[Control^]** or **\[Command⌘]** + **F** to open canvas search. Type to find nodes by their title or display name. Matching nodes are highlighted on the canvas, making it easy to locate specific nodes in large workflows. # Collaboration & Sharing ## Real-Time Collaboration FLORA is optimized for team collaboration. Our canvas supports multiple team members to work on the same design file simultaneously. Changes are visible in real time, ensuring that everyone stays on the same page without version conflicts. ## Annotation Tool FLORA allows team members to leave comments on the canvas, creating a seamless real-time collaboration on the most updated version of your team's creative process. Learn how to put a comment on the canvas [here](/editor/toolbar#comment).
## Share You can **share your project** via copying the link or you can **publish your project** to the community page. ### Share a link To copy the link, navigate to the parameter bar in the top right corner of the canvas and click the **Copy Link** button.
### Publish to Community To publish to community, navigate to the parameter bar in the top right corner of the canvas and click **Publish to Community** button. This makes your project visible to all FLORA users on the Community tab.
# FAUNA FAUNA is an AI agent that lives inside the FLORA canvas. Instead of manually dragging nodes, picking models, writing prompts, and building workflows - you describe what you want and FAUNA builds it. It's not a chatbot. It has hands. It reads your canvas, adds nodes, chooses models, connects pipelines, runs generations, and organizes the output - all from a conversation. {% embed url="" %} *** ### Getting Started Open FAUNA by pressing `⌘ /` or `Ctrl /` anywhere on your canvas. A sidebar will appear on the right side, ready for your prompt. You can also open FAUNA from the FAUNA button in the bottom-right corner of your canvas. {% hint style="info" %} **Quick tip:** Select nodes on your canvas before prompting - they'll automatically be included as context for FAUNA. You can also type `@` in the chat to mention specific nodes by name. {% endhint %} *** ### What FAUNA Can Do #### Chat interface FAUNA is a full conversational agent - not a one-shot prompt box. You can have back-and-forth conversations, iterate on ideas, and build up complex workflows through dialogue. {% embed url="" %} #### Switch modes FAUNA supports two modes of operation, selectable from the mode selector in the chat input: * **Assist mode** (default): FAUNA asks for your explicit confirmation before running any generations. You'll see which nodes it plans to execute and the estimated usage cost before anything runs. * **Auto mode**: FAUNA auto-runs all generations without pausing for approval. Best for when you want maximum speed. #### Generate workflows from scratch Describe what you want to create, and FAUNA will build the workflow for you - selecting appropriate models, configuring nodes, and connecting everything together. **Try prompts like:** * "Generate a minimalist logo for a streetwear brand, create 3 campaign poster variations, then mock up the best one on a billboard and animate it into a 5-second video." * "Storyboard a 4-frame car commercial driving through a desert at sunset, then animate the hero frame into video." * "Generate a cinematic portrait, then create 4 art direction variations - film noir, golden hour, cyberpunk neon, and vintage Polaroid." #### Starter prompts When you first open FAUNA, you'll see quick action buttons to help you get started: * **What's new in FLORA** - Get up-to-date information about the latest FLORA features * **How does FLORA work?** - Learn about FLORA's capabilities * **What are the best models?** - Get model recommendations for your use case * **Create a starter workflow** - FAUNA generates a multi-step workflow you can customize #### Work with your existing canvas Select any nodes on your canvas, and FAUNA can: * Add new nodes that connect to your selection — including generation nodes, action nodes, and layer editor nodes * Modify settings, labels, prompts, models, or aspect ratios * Reconnect or restructure your workflow * Remove nodes individually or by type * Create variations of existing nodes * Extend your workflow with additional steps * Group related nodes into containers * Arrange and organize nodes on your canvas * Use the Layer Editor — FAUNA can operate layer editor tools directly, adding text layers, shapes, and adjustments to Layer Editor nodes on your canvas **Try prompts like:** * "Create 5 variants of this image" * "Change all the image nodes to use Ideogram 3.0" * "Connect these outputs to a new video node" * "Add an upscaling step after each image" * "Remove all the empty nodes" * "Group these image nodes together as 'Hero Shots'" * "Add a text overlay that says 'Summer Sale' to this layer editor" #### Upload images You can attach images directly in the FAUNA chat to use as reference or source material. FAUNA supports: * **Drag and drop** images into the chat input * **Paste** images from your clipboard * **Click the attachment button** to select files Accepted formats: JPEG, PNG, WebP, and GIF (up to 20 MB per file, maximum 5 attachments per message). FAUNA can place attached images onto your canvas as source nodes for other generation nodes to use. #### Message length limit Each message can be up to **25,000 characters**. A character counter appears near the input when you're past 90% of the limit, turning red as you approach the cap. The Send button is disabled once you exceed 25,000 characters. This limit accommodates large prompts and shot lists while keeping conversation context manageable. #### Run your work FAUNA can execute image, video, and text nodes - up to 50 at a time. Ask it to run specific nodes or entire sections of your workflow. {% hint style="info" %} **Approval required (in Assist mode):** When FAUNA wants to run nodes, it will show you which nodes it plans to execute and the estimated usage cost. You must approve before any generation runs - FAUNA won't spend your usage without your explicit confirmation. FAUNA itself is free on every plan and never counts against your budget; the cost confirmation applies only to the nodes FAUNA proposes to run. In Auto mode, FAUNA runs generations immediately. {% endhint %} #### Context about FLORA FAUNA has access to FLORA's documentation and model library. It knows every node type, every model, prompting best practices, and cost/quality tradeoffs. When you ask it to build or run something, it looks up: * How specific nodes work * Which models are available and what they're good at * Best practices for prompting different models * What's new in FLORA *** ### The Sidebar FAUNA lives in a sidebar on the right side of your canvas. It has two layout modes: * **Floating** - The sidebar overlays your canvas as a panel in the bottom-right corner * **Docked** - The sidebar attaches to the right edge of your screen at full height Toggle between docked and floating with `⌘ \` (Mac) or `Ctrl \`. *** ### Keyboard Shortcuts | Shortcut | Action | | ------------------------------- | --------------------------------------------------------------------- | | `⌘ /` (Mac) or `Ctrl /` | Toggle FAUNA sidebar open/close | | `⌘ \` (Mac) or `Ctrl \` | Toggle FAUNA sidebar between docked (right edge) and floating | | `Escape` | Cancel pending execution if there is one, otherwise close the sidebar | | `⌘ Enter` (Mac) or `Ctrl Enter` | When there's a pending execution, approve/continue | | `Enter` | Send message | | `Shift Enter` | New line in chat input | | `@` | Open node picker to mention a specific canvas node | {% hint style="info" %} Modifier keys are platform-aware - Mac shows `⌘` and Windows/Linux shows `Ctrl`. {% endhint %} *** ### Providing Context There are three ways to give FAUNA context about what you're working with: #### 1. Select nodes on your canvas When you select nodes on the canvas, they're automatically included as context in your next message. This is the easiest and most common way to give FAUNA context. #### 2. Use @ mentions Type `@` in the chat input to open a node picker. Search by node name or type, then select the node you want to reference. This is useful when you want to reference a specific node without selecting it on the canvas. #### 3. Upload images Attach images via drag-and-drop, paste, or the attachment button to use as reference material or source images. {% hint style="info" %} You can include up to 15 nodes as context in a single message. {% endhint %} *** ### Thinking & Reasoning FAUNA shows you its thinking process as it works. While FAUNA is processing your request, you'll see: * **A real-time shimmer indicator** showing what FAUNA is currently doing (e.g., "Reading canvas...", "Adding nodes...") * **An expandable thinking timeline** - after FAUNA responds, click "View steps" to see the full reasoning process, including which tools it used and what decisions it made This transparency helps you understand FAUNA's approach and makes it easier to course-correct if needed. *** ### Agentic Collaboration FAUNA works as an agent that does the work for you on canvas, not as a chat assistant. Think of it as a creative co-pilot that operates the tool for you - the closest analogy is Cursor for code. #### Real-time streaming As FAUNA thinks through your request, you'll see its responses stream in real-time. You can watch as it reasons through the problem, decides which tools to use, and takes action on your canvas. #### Cancel anytime Changed your mind? You can cancel FAUNA's response mid-stream if you see it going in a direction you don't want. Your conversation context is preserved, so you can immediately try a different approach. #### Give feedback You can rate FAUNA's responses with thumbs up or thumbs down on any assistant message. This feedback helps us improve FAUNA's performance over time. *** ### Memory & Sessions FAUNA remembers your conversation history within each project. When you return to a project, you can pick up where you left off. #### How memory works * **Per-project sessions:** Each project has its own FAUNA session with its own conversation history * **Persistent messages:** Your messages and FAUNA's responses are saved automatically * **Context preserved:** The nodes you referenced in previous messages are tracked alongside the conversation * **Session history:** Access previous chat sessions from the dropdown in the sidebar header * **Rename sessions:** Give your sessions descriptive names by renaming them from the session dropdown #### Starting fresh If you want to start a new conversation without previous context, you can begin a new session using the new chat button in the top right corner of the sidebar header. Your previous session history remains accessible from the session dropdown. *** ### Canvas Awareness FAUNA understands your canvas in detail and tracks changes between conversation turns. #### Change detection FAUNA automatically detects when your canvas has changed since its last response. When you add, remove, or modify nodes between messages, FAUNA is aware of the changes and factors them into its next response — no need to re-explain what you've done. #### Smart node prioritization When reading your canvas, FAUNA prioritizes the most relevant nodes. Newly added and recently modified nodes are surfaced first with `[NEW]` and `[MODIFIED]` annotations, so FAUNA focuses on what's changed rather than treating all nodes equally. #### Compact node references FAUNA uses short IDs (`n1`, `n2`, `n3`...) instead of long UUIDs when referencing nodes. These IDs are persistent across your session and make it easier to follow which nodes FAUNA is talking about. You can use these same short IDs when referring to nodes in your messages. #### Canvas diagram tool FAUNA can internally visualize your canvas as a graph diagram to understand how nodes are connected, what data types flow between them, and which disconnected workflows exist. This helps FAUNA reason about complex multi-node setups and make better decisions about where to add or modify nodes. *** ### Safety & Reliability FAUNA includes mechanical safeguards that prevent it from making mistakes on your canvas. #### Harness-level enforcement Before modifying any node, FAUNA must first inspect it to understand its current state. This prevents FAUNA from making changes based on stale or outdated information — for example, if you've manually updated a node's prompt since FAUNA last read it, FAUNA will re-inspect the node before editing it. This "look before you touch" approach is enforced at the system level, not just via prompting. #### Telemetry & observability FAUNA's tool usage and decision-making is instrumented with a unified telemetry system, enabling the team to monitor performance, detect issues, and continuously improve FAUNA's reliability. *** ### Current Limitations FAUNA is powerful but still has boundaries. Here's what to expect: #### Single-project scope FAUNA can only see and work with the project you currently have open. It can't access other projects, workspaces, or external websites. #### No access to generation history or past assets FAUNA doesn't yet have access to your existing project context, generation history, or assets from previous sessions. *** ### Coming Soon * **Generate across projects** - Pull assets and context from other projects, not just the one you have open. * **Pull from your context** - FAUNA will have access to your existing project context, generation history, and past assets. * **Multiple FAUNA agents** - Run multiple agents and threads at the same time, so you can direct a team of creatives working in parallel. *** ### Tips for Better Results {% hint style="success" %} **Start small, then scale up.** Begin with simple tasks - changing a model, editing a prompt, or creating a single node. As you get comfortable, try larger requests like building workflow sections or running batch operations. This helps you learn FAUNA's strengths and discover how far you can push it. {% endhint %} **Be specific about your creative intent.** "Generate a cinematic product shot workflow for luxury watches" will get better results than "make something cool." **Select relevant context.** Select nodes on your canvas before prompting, or use `@` to mention specific nodes. The more relevant context you provide, the better FAUNA understands what you're trying to do. **Upload reference images.** Drag and drop or paste images directly into the chat to give FAUNA visual reference material for your creative direction. **Iterate in conversation.** FAUNA works well when you build on previous responses. Start with a base workflow, then refine: "Now add a step that upscales the final output" or "Change the style to be more minimal." **Review before running.** When in Assist mode, take a moment to review the node list and usage estimate before approving. This helps you stay in control of your usage. **Reference nodes.** FAUNA can find nodes by their label, type, content, or model. For example: "make my image nodes use Ideogram 3.0" or "change all the Flux nodes to landscape format." **Correct course when needed.** If you see FAUNA heading in the wrong direction, press Escape to stop it mid-response. Then tell it what went wrong and how to adjust - FAUNA learns from your feedback within the conversation. **Start fresh when switching context.** When you're moving to a completely different part of your project or changing topics entirely, use the new chat button in the sidebar header. This clears the conversation history so FAUNA isn't confused by unrelated context from your previous work. *** ### We Want Your Feedback We're always learning how people want to use FAUNA. What's working? What's frustrating? What do you wish it could do? **To share feedback:** * Rate responses with thumbs up/down directly in the chat * Fill out this short survey: Specific examples help us most - if something didn't work as expected, tell us what you tried and what happened. If something worked great, we want to hear that too. *** ### Troubleshooting **FAUNA isn't responding.** Try closing and reopening the sidebar. If the issue persists, refresh your browser. **FAUNA seems confused about my canvas.** Make sure you've selected the specific nodes you want it to work with, or use `@` mentions to reference them. FAUNA works best with explicit context rather than trying to understand your entire canvas at once. **FAUNA made changes I didn't want.** Use undo (Cmd/Ctrl + Z) to reverse any canvas changes FAUNA made. You can also cancel a response mid-stream if you see it going in the wrong direction. **Ran out of usage during execution.** FAUNA will notify you if your monthly budget runs out. Either upgrade your plan, enable on-demand spending, or wait for your next billing cycle — FAUNA will remember what you were working on. *** *Last updated: June 2026* # Elements ## Summary Elements are reusable references that you can create, manage, and use across your canvas. Elements take one or more images of a single subject — such as a product, character, or object — and create high-quality references that drive more consistent results with less effort. Instead of attaching your raw images directly, Flora studies your inputs and generates a unified **reference** that captures the key features and intent while filtering out unnecessary details. For example, an Element of a person might capture their face and hair while ignoring a specific outfit, pose, or background. The result is a high-signal reference that you can reuse across your canvas to keep a subject consistent from one generation to the next. ## How Elements Work When you create or edit an Element, you provide a few **inputs**: * **Images** — reference images of your subject. * **Category** — what kind of thing the Element is (a model, product, style, and so on), which helps Flora focus on the details that matter for that type of subject. * **Guidelines** *(optional)* — a short text description that highlights important aspects or how the Element should be used. Flora combines these inputs and generates an **Element Preview**: a single reference that represents what your generations will focus on. The preview is your signal for how the Element will behave — if it captures the right details, downstream generations will too. {% hint style="info" %} The Element Preview is a preview of what your generations will focus on. If it's missing detail or includes something it shouldn't, adjust your inputs and regenerate (see [Tips for Better Elements](#tips-for-better-elements)). {% endhint %} {% hint style="warning" %} AI can make mistakes, and generation results will vary. Elements improve consistency, but they can't guarantee perfect consistency or eliminate mistakes. See [Tips for Better Elements](#tips-for-better-elements) for ways to get better results. {% endhint %} ## Creating an Element There are three ways to create a new Element: ### From the Element Library 1. Click the **#** (Elements) button in the toolbar to open the Element Library. 2. Click **Create New Element** or the **New Element** card. 3. Add one or more reference images of your subject (supported formats: PNG, JPEG, WebP, SVG). 4. Choose a **Category** and, optionally, add **Guidelines**. 5. Give your Element a name. 6. Flora generates an **Element Preview** from your inputs. Once you're happy with it, click **Create**. ### From the Canvas (Right-Click) You can create an Element directly from existing image outputs on the canvas: 1. Select one or more image nodes that have generated outputs. 2. Right-click to open the context menu. 3. Select **Create Element**. Flora uses the selected image outputs as the inputs for the new Element. ### From an Element Node 1. Add an Element node to the canvas (from the toolbar or by dragging from the Element Library). 2. Open the node's picker and select **New Element** at the bottom of the list. ## The Element Editor The Element editor is where you set an Element's inputs and shape its reference. ### Inputs * **Images** — add up to 8 images of the same subject, varying angles, lighting, and framing. * **Category** — pick the category that best describes your subject. Categories include **Model**, **Character**, **Product**, **Object**, **Material**, **Setting**, **Style**, **Moodboard**, and **General**. The category helps Flora push for the right kind of consistency (for example, facial and physical features for a model, or visual fidelity and accurate details for a product). * **Guidelines** *(optional)* — describe how the Element should be used or call out details that matter. Guidelines help you add context alongside your images. ### Element Preview As you set your inputs, Flora generates the **Element Preview** — the reference image that drives your generations. * **Regenerate** — generate a fresh reference from the current inputs. Use this if the preview is missing detail or includes something it shouldn't. * **Set as Default** — when you've generated more than one reference, choose which one the Element uses. * **Reference history** — every reference you generate is saved to a thumbnail strip below the preview. Click any past reference to view it, then **Set as Default** to revert to it. This lets you keep generating and pick the best result. ## Tips for Better Elements Elements turn your inputs into a high-signal reference, so the quality of your inputs matters most. * **Better inputs produce better results — and more isn't always better.** Stick to **5 or fewer images** when possible. A few clear, varied images of the same subject usually beat a large, noisy set. * **Vary angles, lighting, and framing.** Close-ups, wide shots, and different perspectives all help Flora understand the subject. * **Add Guidelines to highlight what matters.** Optionally use the Guidelines field to call out important aspects of the subject. * **If a reference has missing or incorrect detail, regenerate or adjust your inputs.** Try regenerating the Element, or change your images, category, or guidelines until you get a high-quality result. You can always revert to a past version to select the best reference. ## Using Elements ### The Element Node The Element node acts as a visual input reference on the canvas. When you add an Element, the node shows the Element's reference, name, and category. To change which Element a node references, open the node's picker and search or browse — Elements are grouped by category — then select one. ### Mentioning Elements in Prompts You can also reference an Element directly inside a prompt by mentioning it. The picker groups Elements by category so they're easy to find. ### Connecting to Image Nodes Element nodes have an output handle you can connect to compatible image generation nodes. When connected, the model uses the Element's reference to maintain the subject's visual identity in the generated output. Elements work with image-to-image and image-set-to-image models, including: * Nano Banana Pro * Nano Banana 2 * GPT Image 2 * Seedream 4.5 * Riverflow 2.0 Fast and Pro {% hint style="success" %} Some models are much better at driving visual consistency than others. For images, we recommend **Nano Banana Pro**, **Nano Banana 2**, or **GPT Image 2** for the best results. {% endhint %} ### Connecting to Video Nodes Elements can also be connected to compatible video generation models to create videos that maintain your subject's visual identity. Supported video models include: * Seedance 2.0 (reference mode) * Kling O1 * Kling O3 Connect an Element node's output to a supported video node just as you would with an image node. {% hint style="success" %} For videos, we recommend **Seedance 2 References** to get the best results. {% endhint %} ### Drag and Drop You can drag an Element card from the Element Library directly onto the canvas. This creates an Element node at the drop location, ready to connect to other nodes. ## Managing Elements ### The Element Library Open the Element Library by clicking the **#** button in the toolbar. From here you can: * **Switch** between the **Private** tab (your own Elements) and the **Shared** tab (Elements shared with your workspace). * **Search** for Elements by name using the search bar. * **Sort** Elements by name (A to Z, Z to A) or by date (Newest, Oldest). * **Share** an Element to your workspace (or unshare it) via the right-click context menu. * **Edit** any Element by hovering over its card and clicking **Edit element**, or via the right-click context menu, to open the Element editor. * **Add to canvas** by clicking **Add to canvas** on the card, or by dragging the Element card directly onto the canvas. * **Delete** an Element via the right-click context menu. Deleting an Element unbinds it from any canvas nodes that reference it. ### Editing an Element To change an Element's inputs, reference, name, or category: 1. Open the Element Library. 2. Hover over the Element card and click **Edit element** (or right-click → **Edit element**). 3. In the Element editor, update the images, category, guidelines, or name, and regenerate the reference as needed. 4. Click **Update** to save your changes. ## Tips * Elements are **private** to you by default, but you can **share** an Element to your workspace so others can use it. Use the **Private** and **Shared** tabs in the Element Library to switch between your own Elements and those shared with the workspace, and the right-click menu to **Share to workspace** or **Unshare from workspace**. * You can create Elements from any image output on the canvas, making it easy to turn a generation you like into a reusable reference. * The Element Preview is a preview of what your generations will focus on — treat it as your signal for quality. * If an Element is deleted while it's still in use on the canvas, the node shows an "Element unavailable" state. ## Troubleshooting ### My reference doesn't look like my input AI can occasionally miss details. Try regenerating the reference, and add **Guidelines** describing the specific details you want to capture. Also make sure the Element's **Category** is correct, since it shapes what Flora focuses on. ### Using an Element produces duplicates in my output Be specific about what you want in your prompt. Describe the scene and how the Elements relate, rather than just listing them. For example, "product shot of #model holding #product" works better than "#model with #product". ## Limits and Fallback Behavior A few things to keep in mind about reference generation: * **AI can make mistakes, and results will vary.** Elements improve consistency, but they can't guarantee perfect consistency or eliminate mistakes across generations. See [Tips for Better Elements](#tips-for-better-elements) for ways to get better results. * **Reference generation may depend on available models.** If models are disabled for your workspace, Flora may not be able to generate a reference for an Element. * **Reference generation may be rate limited.** If you regenerate frequently, you may need to wait before generating another reference. * **Elements without a reference fall back to their images.** If an Element doesn't have a generated reference, Flora forwards the Element's images directly as an attachment instead. You'll still get a result, but without the high-signal reference that drives the most consistent output. # Fashion Studio A focused workspace for fashion teams — turn sketches into renders, swap colors and garments, and review everything in one feed. ## Overview Fashion Studio is a purpose-built workspace for fashion teams. Instead of the full FLORA canvas, it gives you a curated set of garment tools in a simple three-panel interface: pick a tool, drop in your assets, and generate — no node wiring required. It's designed for the everyday work of a fashion team: turning concept sketches into production-grade renders, recoloring garments across a catalog, dressing models in new looks, and reviewing all the results together in one place. {% hint style="info" %} Fashion Studio is the first of FLORA's **Studios** — focused creative environments built on top of the canvas. Everything you make in a studio can be opened in the full canvas for deeper editing. {% endhint %} ## Getting Started 1. Open **Studios** from your dashboard. 2. Click the **Fashion Studio** card. This opens your most recent studio project, or creates one if it's your first visit. (Use the **⋯** menu on the card to explicitly open the last edited project or start a new one.) 3. The first time you enter, a short **Introducing Fashion Studio** walkthrough shows you around. Dismiss it anytime — the studio is live underneath. ## The Interface The studio has three panels, left to right: ### Tool rail and settings The left sidebar lists every tool in the studio. Click a tool to select it, then fill in its inputs in the settings panel — garment images, sketches, colors, text, and reference photos, depending on the tool. Some inputs accept **collections**, so you can batch one generation across many items at once (for example, one garment in every colorway). The footer shows the credit cost for the run. When your inputs are ready, hit **Generate**. {% hint style="info" %} If a tool shows a **New version available** banner, click **Update** to get the latest version of that tool. {% endhint %} ### The feed The center panel is your results feed. Every generation lands here, grouped by run (one run per click of Generate), newest first, with a timestamp and the tool that produced it. * **View controls**: Switch between a grid and a horizontal ticker layout, and adjust thumbnail size with the slider. * **Per-result actions**: Hover a thumbnail to **Download** or open it **fullscreen**. The **⋮** menu adds **Show info**, **Open in Canvas**, and **Delete**. * **Batch actions**: Select multiple results with Shift+click or Cmd/Ctrl+click, then download them all or send them to the canvas together. * **Run actions**: Download or delete an entire run from its header. ### Details panel Click the info toggle in the top-right (or **Show info** on a result) to slide open the details panel. It shows the selected result's type, resolution, file size, name, model, and generation time, with a download button. ## The Tools Fashion Studio ships with four tools: | Tool | What it does | | ---------------------- | ---------------------------------------------------------------------------- | | **Sketch to Render** | Turn a flat sketch or technical drawing into a photorealistic garment render | | **Garment Extractor** | Isolate a garment from a photo as a clean standalone asset | | **Garment Color Swap** | Recolor a garment while preserving fabric, texture, and construction details | | **Garment Swap** | Swap a garment onto a model, keeping the fit and look consistent | ### Coming soon The tool rail also previews what's on the roadmap. These appear dimmed until they ship: * **Garment Refinement** — Sharpen fit, fabric, and detail on an existing garment render. * **Change Pose** — Restage the same look in a new pose. * **Multi-Angle Shoot** — Reshoot the same look from new camera angles. * **Pattern Placement** — Drop a logo, badge, or patch onto a garment, depth-warped to the fabric. * **Print Generator** — Generate original prints with tiling and repeat controls. * **Pattern Swap** — Wrap a print across a whole garment, depth-warped to its fit and drape. ## Projects Your work in Fashion Studio is organized into projects, just like the canvas. * The breadcrumb at the top of the feed shows **Fashion Studio / Project name**. Click it to search and switch between projects, rename the current one, or start a **New project**. * Each project keeps its own feed history, so you can dedicate projects to different collections, drops, or clients. * Studio projects also appear in your dashboard project list with a **Studio** badge. ## From Studio to Canvas Studios are powered by the FLORA canvas, so you're never boxed in: * **Open in Canvas** on any result seeds it into a canvas project as an image node, ready for further editing, upscaling, or animation. * **Send to Canvas** on a multi-selection brings a whole batch over at once. Anything the studio can't do, the full canvas can — see the [Canvas](/editor/canvas) and [Techniques](/nodes/techniques) docs to go deeper. ## Working with Your Team Fashion Studio projects support the same collaboration as the rest of FLORA. Invite your team into the workspace and review renders together in the feed — design, e-comm, and brand in the same room before anything ships. Collaborators with edit access can generate with any tool; deleting runs and results is reserved for the project owner. ## Common Questions **How much does a generation cost?**\ Each tool shows its credit cost next to the Generate button before you run it. Costs vary by tool and settings, and come out of your workspace's regular credit balance — studio generations aren't priced differently from canvas generations. **Can I batch across a whole catalog?**\ Yes — tools with collection inputs accept multiple items in one run, so you can generate a garment across many colorways, swatches, or models in a single click. **Where do my results live?**\ Everything stays in the project's feed until you delete it. Download results individually, per run, or by selection — or send them to the canvas to keep working. **I don't see Studios in my dashboard.**\ Studios are rolling out gradually. If you don't see the Studios tab yet, it hasn't reached your account — check back soon. # Text Models Here are all the text models currently available in FLORA.
ModelEst. time (s)DescriptionBest for
Claude Fable 525Anthropic's model capable of the most demanding reasoning and long-horizon agentic work.Anthropic's model capable of the most demanding reasoning and long-horizon agentic work.
Claude Opus 4.525Combines maximum capability with practical performance.In-depth domain expertise, nuanced dialogue, legal or policy analysis, and strategic decision support.
Claude Opus 4.68Anthropic's frontier model, supports adaptive thinking.Anthropic's frontier model, supports adaptive thinking.
Claude Opus 4.725Flagship model by Anthropic that handles complex, long-running tasks.Most advanced Anthropic reasoning, complex analysis, and highest-fidelity outputs.
Claude Opus 4.825Anthropic's most capable model for complex reasoning.Anthropic's most capable model for complex reasoning.
Claude Opus 525Anthropic's most capable model for complex reasoning.Anthropic's most capable model for complex reasoning.
Claude Sonnet 4.525Balanced speed and intelligence.Balanced performance, creative writing, code generation, and versatile task handling.
Claude Sonnet 4.68State-of-the-art results at fast speeds.Balanced performance, creative writing, code generation, and versatile task handling.
Claude Sonnet 535State-of-the-art results at fast speeds.State-of-the-art results at fast speeds.
Gemini 2.5 Pro30Google's best for complex reasoning.Enterprise-scale analytics, multimodal Q&A, detailed data synthesis, and robust safety-critical tasks.
Gemini 3 Flash25Transcribe audio files with reasoning.Instant summarization, real-time assistance, mobile-first interactions, and search tasks.
Gemini 3.1 Pro45Google's latest and most capable reasoning model.Advanced reasoning, complex analysis, and Google's most capable language model.
GPT 4o Mini20Fast and affordable for simple tasks.Quick drafts, instant summarization, lightweight chatbots, and real-time user feedback loops.
GPT-540Advanced reasoning with deep thinking.Advanced reasoning, long-context analysis, high-precision code generation, and complex multi-step problem-solving.
GPT-5.130OpenAI's advanced reasoning model.Enhanced reasoning, complex analysis, and multi-step problem-solving with improved accuracy.
GPT-5.235OpenAI’s model marking significant improvements in general intelligence.Cutting-edge reasoning, highest accuracy, and most capable OpenAI model.
GPT-5.435Versatile text model with advances in reasoning.State-of-the-art reasoning, maximum accuracy, and OpenAI's flagship model.
GPT-5.524Frontier model suitable for complex professional work.Frontier model suitable for complex professional work.
o3 Deep Research600Use multi-step analysis with web search support.In-depth research, comprehensive analysis, multi-source synthesis, and complex investigation tasks.
# Claude Sonnet 4.5 (by Anthropic) ### Overview Anthropic's strongest balanced model for versatile task handling. *** ### Quick facts * **Modes:** Text → Text · Image → Text · Video → Text * **Default output / size:** Text responses (token-limited) *** ### What it's great for * Balanced performance across all task types. * Creative writing and content generation. * Code generation and technical tasks. * Versatile handling of diverse requests. *** ### Parameters | Control | Type | Default | Notes | | ----------- | ---- | ------- | --------------------------------------- | | Prompt | Text | — | Natural-language instruction (required) | | Input Texts | List | — | Optional supporting documents | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ---------------------------- | | Text → Text | \~8s | Prompt | Balanced reasoning & writing | | Image → Text | \~8s | Images, Prompt | Visual analysis | | Video → Text | \~8s | Videos, Prompt | Video understanding | *** ### Prompt tips * Great all-around model for diverse tasks. * Balance of speed and capability. * Ideal when you need versatility without sacrificing quality. *** ### Safety Moderation is enforced under Anthropic policies. *** # Claude Opus 4.5 (by Anthropic) ### Overview Anthropic's most intelligent model with advanced reasoning capabilities. *** ### Quick facts * **Modes:** Text → Text · Image → Text · Video → Text * **Default output / size:** Text responses (token-limited) *** ### What it's great for * Most advanced Anthropic reasoning. * Complex analysis and deep understanding. * Highest-fidelity outputs for demanding tasks. * Expert-level domain knowledge. *** ### Parameters | Control | Type | Default | Notes | | ----------- | ---- | ------- | --------------------------------------- | | Prompt | Text | — | Natural-language instruction (required) | | Input Texts | List | — | Optional supporting documents | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ----------------------------- | | Text → Text | \~28s | Prompt | Advanced reasoning & analysis | | Image → Text | \~28s | Images, Prompt | Deep visual analysis | | Video → Text | \~28s | Videos, Prompt | Comprehensive video summaries | *** ### Prompt tips * Best for tasks requiring maximum intelligence and nuance. * Preload context via Input Texts for grounded reasoning. * Ideal for strategic, analytical, and expert-level tasks. *** ### Safety Moderation is enforced under Anthropic policies. *** # GPT-5 (by OpenAI) ## **GPT-5 (OpenAI)** **Advanced multimodal language model for complex reasoning, orchestration, and multimodal tasks.** *** ### **Quick facts** * **Modes:** Text → Text · Image → Text · Video → Text * **Default output / size:** Text responses (token-limited) *** ### **What it’s great for** * Complex coding, reasoning, and orchestration tasks. * Video → text transcription & summarization. * Multimodal pipeline orchestration. *** #### **Copy-and-paste prompts** ``` Summarize this 30s product demo video into a 3-bullet marketing blurb. ``` ``` Extract timestamps and scene descriptions from this video. ``` *** ### **Parameters** | Name | Type | Default | Notes | | -------- | ---- | ------- | -------- | | `Prompt` | Text | — | Required | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------- | --------------- | ------------------------------- | | Text → Text | Streaming | Prompt | Code, summaries, long-form text | | Image → Text | Streaming | Images | Image description | | Video → Text | Streaming | Videos, Prompt | Transcription & summarization | *** ### **Output options** | Option | Values / notes | | ------- | ----------------------------------------------------------- | | Formats | Text / JSON | | Notes | Advanced reasoning & tooling integrations available via API | *** ### **Prompt tips** * Be explicit about length and format (bullets, code block, JSON). * Provide context and any relevant assets. *** # GPT-5.1 (by OpenAI) ### Overview OpenAI's advanced multimodal language model with enhanced reasoning capabilities. *** ### Quick facts * **Modes:** Text → Text · Image → Text · Video → Text * **Default output / size:** Text responses (token-limited) *** ### What it's great for * Enhanced reasoning and complex analysis. * Multi-step problem-solving with improved accuracy. * Multimodal understanding across text, images, and video. * Code generation and technical documentation. *** ### Parameters | Control | Type | Default | Notes | | ------- | ---- | ------- | --------------------------------------- | | Prompt | Text | — | Natural-language instruction (required) | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ----------------------------- | | Text → Text | Streaming | Prompt | Reasoning, code, long-form | | Image → Text | Streaming | Images, Prompt | Image analysis | | Video → Text | Streaming | Videos, Prompt | Transcription & summarization | *** ### Prompt tips * Be explicit about length and format (bullets, code block, JSON). * Provide context and relevant assets for best results. * Use for complex multi-step reasoning tasks. *** # GPT-5.2 (by OpenAI) ### Overview OpenAI's latest and most advanced multimodal language model. *** ### Quick facts * **Modes:** Text → Text · Image → Text · Video → Text * **Default output / size:** Text responses (token-limited) *** ### What it's great for * Cutting-edge reasoning with highest accuracy. * Most capable OpenAI model for complex tasks. * Advanced multimodal understanding. * Premium quality outputs for professional use. *** ### Parameters | Control | Type | Default | Notes | | ------- | ---- | ------- | --------------------------------------- | | Prompt | Text | — | Natural-language instruction (required) | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ---------------------------- | | Text → Text | Streaming | Prompt | Premium reasoning & analysis | | Image → Text | Streaming | Images, Prompt | Advanced image understanding | | Video → Text | Streaming | Videos, Prompt | Detailed video analysis | *** ### Prompt tips * Best for tasks requiring maximum accuracy and capability. * Ideal for professional and enterprise use cases. * Supports extended reasoning for complex problems. *** # o3 Deep Research (by OpenAI) ### Overview Advanced research model with multi-step analysis and web search capabilities. *** ### Quick facts * **Modes:** Text → Text · Image → Text · Video → Text * **Default output / size:** Extended text responses (up to 180,000 tokens) *** ### What it's great for * In-depth research requiring multiple sources. * Comprehensive analysis with web search integration. * Multi-source synthesis and fact-checking. * Complex investigation and discovery tasks. *** ### Parameters | Control | Type | Default | Notes | | ------- | ---- | ------- | --------------------------------------- | | Prompt | Text | — | Natural-language instruction (required) | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ---------------------------- | | Text → Text | \~2 min | Prompt | Deep research & analysis | | Image → Text | \~2 min | Images, Prompt | Visual research with context | | Video → Text | \~2 min | Videos, Prompt | Video-based research | *** ### Prompt tips * Best for research tasks requiring thorough investigation. * Model will search the web to gather comprehensive information. * Ideal for synthesizing information from multiple sources. * Allow extra time - deep research is computationally intensive. *** # Image Models Here are all the image models currently available in FLORA. ## Generation Models These models create new images from text prompts.
ModelEst. time (s)DescriptionBest for
Flux 230High-resolution output, great for print.Fast, customizable generation with acceleration options
Flux 2 Flex45Adjustable quality/speed balance for production.Consistent outputs across generations, reliable results
Flux 2 Klein 4B25Precise modifications using natural language descriptions.Targeted edits, natural language-driven image changes
Flux 2 Klein 9B50Enhanced realism, crisper text and better color adherence.Realistic outputs with accurate text and color rendering
Flux 2 Max60Exceptional realism, precision, and consistency.Premium photoreal outputs, maximum quality generation
Flux 2 Pro45Top-tier Flux quality with multi-reference support.Professional-grade generation, best BFL quality
Flux 2 Turbo25Speed-optimized Flux 2 model.Fast iteration, quick previews, time-sensitive workflows
Flux Dev20Quick drafts and rapid iteration.Rapid prototyping, mood-boards, style fusion, abstract compositions
Flux Pro 1.124Ultra-high resolution for large-format work.Professional-grade generation, premium quality outputs
GPT Image80Great prompt comprehension, natural language edits.Narrative-driven visuals, branding mock-ups, precise in-painting & style transfer
GPT Image 1.575High-fidelity images with strong prompt adherence.Premium quality generation, transparent backgrounds, precise control
GPT Image 2220Flagship image model from OpenAI.High-fidelity image generation, prompt-accurate visuals, and production-ready assets.
Grok Imagine20Fast image editing model from xAI.Quick generation with xAI's latest image capabilities
Grok Imagine Quality30Quality image editing model from xAI.Quality image editing model from xAI.
Ideogram 3.018Best for logos and art with crisp, readable type.Text-in-image rendering, stylized aesthetics with curated presets
Ideogram 4.018Producing crisp visuals with accurate text rendering.Producing crisp visuals with accurate text rendering.
Imagen 326Realistic edits to existing images.High-quality image editing powered by Google
Imagen 425Photorealistic results with accurate text rendering.High-quality generation with Google's latest image AI
Kling O165Precise edits with text, image, or video guidance.Detailed image editing with high precision
Krea 2 Large30High-fidelity, stylized image model.High-fidelity, stylized image model.
Krea 2 References Large30Generate stylized results using reference images as input.Generate stylized results using reference images as input.
Krea 2 References Medium30Generate stylized results using reference images as input.Generate stylized results using reference images as input.
Lora Trainer120
Luma Photon35Realistic lighting and rendering for objects and architecture.Versatile generation across diverse styles, fast iteration
Nano Banana50Fast and lightweight, good for iterating.Fast, intuitive photo edits & multi-image blends for social-ready transformations
Nano Banana 280Google's new state-of-the-art image model.Next-gen image quality with Google's latest capabilities
Nano Banana 2 Lite20Google's new state-of-the-art image model.Google's new state-of-the-art image model.
Nano Banana Pro90Highest quality from Google, up to 4K.Premium 4K generation, professional multi-image composition, highest quality outputs
Qwen Image 2.0120Next-gen unified editing model.Next-gen unified editing model.
Recraft V325Clean, vector-style illustrations and graphic design.Concept art and creative storytelling
Recraft V440High-quality image generation with enhanced detail.Premium illustrations, detailed graphics, professional design assets
Recraft V4 Pro60High-quality image generation with enhanced detail.Top-tier design outputs, maximum fidelity illustrations
Recraft V4.130Tuned for brand systems and editorial work.Tuned for brand systems and editorial work.
Recraft V4.1 Pro60Made for hero imagery, campaign work, and print.Made for hero imagery, campaign work, and print.
Recraft V4.1 Utility30Optimized for quick, cost-effective image generation with a design focus.Optimized for quick, cost-effective image generation with a design focus.
Recraft V4.1 Utility Pro30Premium-quality raster generation viable across full creative pipelines.Premium-quality raster generation viable across full creative pipelines.
Reve 2.130Reve 2.1 API ModelReve 2.1 API Model
Riverflow 2.0 Fast150High quality image model with optimized latency.Fast, high-quality generation with low wait times
Riverflow 2.0 Pro245Premium image model with high accuracy.Professional outputs with maximum accuracy
Riverflow 2.5 Pro245Model suitable for commercial visual work that needs stronger output quality.Model suitable for commercial visual work that needs stronger output quality.
Seedream 3.020Cinematic film-style visuals with legible text.Stylized marketing content, cinematic visuals
Seedream 4.050Strong spatial understanding, good for batch work.4K photoreal marketing visuals, consistent product-hero images, educational diagrams with precise labels
Seedream 4.565Sharp text rendering with 4K output.High-quality photoreal images, multi-image composition, 4K asset creation
Seedream 5 Pro30Efficient content creation and production capabilitiesEfficient content creation and production capabilities
Seedream 5.0 Lite30Next-gen image editing with enhanced prompt understanding.Next-gen image editing with enhanced prompt understanding.
Stable Diffusion 3.570Open-source model for general creation and customization.General-purpose image generation and rapid iteration
Uni-130Edit an image using Luma's multimodal reasoning model.Edit an image using Luma's multimodal reasoning model.
Uni-1 Max30Higher-quality unified image model by Luma.Higher-quality unified image model by Luma.
Wan 2.225High realism with fine detail control.Cost-effective realistic image generation
Z-Image Turbo20Fast image-editing model with strength control.Rapid iteration, budget-friendly generation, quick prototyping
## Editing Models These models edit or transform existing images.
ModelEst. time (s)DescriptionBest for
Enhancor V132Portrait skin smoothing and refinement.Subtle portrait retouching, skin smoothing
Enhancor V3185Professional-grade face enhancement.Professional portrait retouching and enhancement
Enhancor V470Professional-grade face enhancement.Advanced face enhancement with latest capabilities
Flux Canny30Extract precise edges to maintain composition.Precise composition control via edge detection for detailed scenes
Flux Depth65Create depth maps for 3D effects and spatial transforms.Maintaining 3D spatial consistency and accurate perspectives
Flux Kontext Max30Complex scenes with strong spatial awareness.Out-painting, style swaps, text edits, maintaining character continuity
Flux Redux28Subtle variations of the input image.High creative freedom with loose compositional guidance
GPT Image 1.5 Inpainting60OpenAI's powerful image inpainting model.Precise inpainting with strong prompt comprehension
Ideogram Character30Transfer faces between images.Maintaining character identity across generations
Imagen 3 Outpainting26Google's powerful image editing model.Extending images beyond their original borders
Nano Banana Pro Inpainting65Highest quality from Google, up to 4K.Premium inpainting with Google's best image model
Qwen Image Edit25Best for adding or changing text in images.Text editing within images, precise inpainting
Qwen Image Edit 2511 Angles24Full 360 camera angle control with horizontal and vertical rotation.Full 360 camera angle control with horizontal and vertical rotation.
Qwen Image Edit Plus120Complex layouts with multiple text elements.Complex text edits, advanced image composition
Remove background15Isolate clean subjects from images.Subject isolation, transparent backgrounds, compositing
Riverflow 2.0 Pro Inpainting210Premium image model with high accuracy.Precise inpainting with high-accuracy results
## Vector & SVG Models These models generate or convert vector graphics.
ModelEst. time (s)DescriptionBest for
Arrow 1.0500Convert raster images to production-ready SVG vectors.Raster-to-vector conversion, production-ready SVG output
Arrow 1.0 References500Text-to-SVG generation guided by up to 4 reference images.Guided SVG creation with style and composition references
Arrow 1.1500Most efficient raster-to-vector model.Faster raster-to-vector conversion with improved structure and quality.
Arrow 1.1 Max500Generate high quality SVG vectors using an image input.Premium vector generation for detailed and complex illustrations.
Arrow 1.1 Max References500Highest quality text-to-SVG generation with reference image guidance.Maximum quality reference-guided SVG generation.
Arrow 1.1 References500Updated and more efficient text-to-SVG model guided by image references.Reference-guided SVG generation with improved speed and consistency.
Recraft V4 Pro Vector55High-quality vector illustration generation.Premium vector illustrations, maximum fidelity
Recraft V4 Vector40Vector illustrations with enhanced detail.Detailed vector illustrations and graphic design
Recraft V4.1 Pro Vector55High-quality vector illustration generation.High-quality vector illustration generation.
Recraft V4.1 Vector40Built for logos, icons, and illustration systemsBuilt for logos, icons, and illustration systems
## Upscaling & Enhancement These models upscale resolution or enhance image quality.
ModelEst. time (s)DescriptionBest for
Magnific Creative Upscaler95Artistic upscaling with creative options.Creative upscaling with stylistic enhancement
Magnific Precision Upscaler65Detail-preserving 2× upscaling.Sharp, faithful upscaling that preserves original detail
Magnific Precision Upscaler V260Sharp edges and texture recovery.Maximum detail recovery, edge-preserving upscaling
Topaz Generative Upscaler60Generative upscaling, reconstructs missing detail, faces, and textures.Generative upscaling, reconstructs missing detail, faces, and textures.
Topaz Upscaler25Boost resolution and detail, industry-standard.Industry-standard upscaling for professional workflows
# Flux Dev (by Black Forest Labs) ### **Overview** Highly customizable text→image model for broad creative use in FLORA. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** Square (1:1) 1024×1024 * **Aspect ratios:** Landscape (16:9), Landscape (4:3), Square (1:1), Portrait (3:4), Portrait (9:16) *** ### **What it’s great for** * Versatile creative generation across styles. * Iterative design and concept exploration. * Fine-grained parameter control in the editor. * Default model for all our presets/ LoRA (styles) *** ### **Example outputs** {% hint style="info" %} POV scene — 35mm shallow depth of field {% endhint %}
{% hint style="info" %} Stylized illustration render {% endhint %}
{% hint style="info" %} Photographic study — editorial portrait {% endhint %}
#### **Copy-and-paste prompts** ``` POV wide-angle shot of a cute lamb on a red-and-white rope lead, nose-to-camera in sharp focus with creamy bokeh, standing in a golden field at sunset with a few sheep softly blurred in the background and streaked clouds overhead, soft warm light, shallow depth of field, 35mm look, high detail. ``` ``` A temple inside a retro-futurist lab with CRT stacks, scanline bloom, vaporwave gradient sky, palm shadows, chrome piping, lens dirt, nostalgic yet high fidelity. ``` ``` Editorial fashion portrait, high contrast, film grain, ``` *** ### **Parameters** | Name | Type | Default | Notes | | ------- | ------ | ------------ | -------------------------------------------------------------------------------- | | Prompt | Text | — | Required description of the scene | | Quality | Select | 100 | Controls render quality (fine-tunes detail) | | Size | Select | Square (1:1) | Landscape (16:9), Landscape (4:3), Square (1:1), Portrait (3:4), Portrait (9:16) | | Seed | Seed | Random | Optional for repeatable runs | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------- | --------------- | ------------------------ | | Text → Image | 4s | Prompt | General image generation | *** ### **Output options** | Option | Values / notes | | ------------ | ------------------------- | | Aspect ratio | 16:9, 4:3, 1:1, 3:4, 9:16 | | Quality | Range is 0-100 | *** ### **Prompt tips** * Use camera / lens tokens for photographic realism. * Increase quality for more refined detail. # GPT-Image (by OpenAI) ### **Overview** Multimodal image generation and editing powered by API. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (auto) * **Aspect ratios:** Auto, 3:2, 1:1, 2:3 *** ### **What it’s great for** * Photoreal and stylized image generation integrated with LLM workflows. * Simple image edits and inpainting. *** ### **Example outputs** {% hint style="info" %} Generated hero image — cinematic portrait {% endhint %}
{% hint style="info" %} Photoreal render {% endhint %}
{% hint style="info" %} Edited photo example — background swap {% endhint %}
#### **Copy-and-paste prompts** ``` Cinematic street portrait, 50mm lens feel, golden hour, shallow depth of field — 1:1 ``` ``` Edit: replace background with moody studio backdrop, natural skin tones preserved ``` ``` Product pack shot on white, crisp shadows, 1200x1200 ``` *** ### **Parameters** | Name | Type | Default | Notes | | -------------- | ------ | ------- | ------------------- | | `Prompt` | Text | — | Required | | `Aspect Ratio` | Select | Auto | Auto, 3:2, 1:1, 2:3 | | `Quality` | Select | High | Low, Medium, High | | `Image` | Upload | — | Required for edits | *** ### **Modes (summary)** | Mode | Estimated time | Required inputs | Typical use | | -------------------- | -------------- | --------------- | ------------------ | | Text → Image | 45 seconds | Prompt | Image generation | | Image → Image (edit) | 1 minute | Image, Prompt | Inpainting & edits | *** ### **Output options** | Option | Values / notes | | ------------ | ------------------- | | Aspect ratio | Auto, 3:2, 1:1, 2:3 | *** ### **Prompt tips** * Specify composition, lens feel, lighting, and desired ratio. * Use “Edit:” to indicate local edits vs full generation. # Nano Banana (Gemini 2.5 Flash Image) (by Google) ## **Overview** High-fidelity image generation and editing (photo-real & stylized) — fast, edit-first workflows. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) · Images → Image. * **Default output / size:** 1024×1024 (Square) * **Aspect ratios:** Square, Landscape, Portrait (determined by input image) *** ### **What it’s great for** * Fast photo edits and stylized generation. * Turning photos into collectible / toy-style renders. * Multi-image fusion and targeted local edits. *** ### **Example outputs** {% hint style="info" %} Hero portrait — edited photo to collectible render {% endhint %}
{% hint style="info" %} Stylized figurine render {% endhint %}
{% hint style="info" %} Multi-image fusion composite {% endhint %}
#### **Copy-and-paste prompts** ``` Portrait photo retouched into collectible 3D figurine on acrylic base, studio lighting, crisp detail ``` ``` Edit: remove background and add dramatic rim light, maintain subject skin tones, photoreal ``` ``` Blend two product images into a single shelf display with natural shadows and reflections ``` *** ### **Parameters** | Name | Type | Default | Notes | | ----------------- | --------- | ------- | ------------------------------------------------ | | `Prompt` | Text | — | Natural-language description (required) | | `Reference Image` | Image | — | Required for edits / multi-image fusion | | `Aspect Ratio` | Inherited | Square | Determined by input reference image aspect ratio | *** ### **Modes (summary)** | Mode | Estimated time | Required inputs | Typical use | | -------------------- | -------------- | ------------------------ | ----------------------------------- | | Text → Image | 20 seconds | Prompt | Generate images from text | | Image → Image (edit) | 30 seconds | Reference Image, Prompt | Local edits, background replacement | | Images → Image | 30 seconds | Reference Images, Prompt | Multi-image fusion / composites | *** ### **Prompt tips** * For edits prefix with “Edit:” and describe pixel-accurate changes. * Use clear style tokens (e.g., “photoreal”, “3D figurine”, “studio lighting”). *** # Seedream 4.0 (by Bytedance) ### **Overview** Consistent, reasoning-driven text→image model with high-resolution outputs. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Outputs / sizes:** 1024×1024 (1:1), 1280×720 (16:9), 720×1280 (9:16) — higher when selecting 2K/4K * **Aspect ratios:** 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 *** ### **What it’s great for** * High-detail photoreal images and editorial art. * High-resolution asset prototyping. * Image editing and inpainting tasks. *** ### **Example outputs** {% hint style="info" %} Shadow study — experimental fashion film still {% endhint %}
{% hint style="info" %} Cinematic character moment {% endhint %}
{% hint style="info" %} Detailed environment concept {% endhint %}
#### **Copy-and-paste prompts** ``` Experimental fashion film still of a blurred shadow silhouette of a figure with long hair reaching out a hand. The body is partially obscured by tree leaf shadows cast on a white wall, captured in soft diffusion. An interior bedroom wall is bathed in hard daylight, with high-contrast backlight forming abstract silhouettes. Shot on 16mm film with a wide lens, extreme close-up low angle composition, desaturated color palette with subtle grain. The mood is hypnotic, dreamlike, and intimate, evoking trance-like disorientation and fragile presence. ``` ``` Cool desaturated digital cinematography, a middle-aged Japanese man in a grey t-shirt brushes his teeth at a stainless steel sink. A towel hangs behind him, toothbrush in mouth, framed in a narrow Tokyo apartment bathroom. Artificial fluorescent side light casts cyan tones across metal surfaces, mixing with warm glow from a frosted glass window. Shot on Sony VENICE 2 with Canon K35 wide lens, medium wide clean single, centered composition, naturalistic production design with water heater and simple domestic objects. The mood is intimate and contemplative, evoking solitude and the quiet poetry of daily ritual. ``` ``` Cinematic landscape film still of furrowed rows of young green crops, vast open field stretching to the horizon. Golden-hour sunset on the right edge throws long shadows and warm amber flare across desaturated earth tones and cool greens. Wide lens, very low ground-level angle with strong leading lines converging toward the sun, deep depth of field. Subtle 16mm grain, gentle halation around the sun, soft vignette. Sky in pastel lavender and peach, horizon thin and distant, mood contemplative, tender, and quietly transcendent, evoking rest after labor and a meditative union with the landscape. ``` *** ### **Parameters (editor-ready)** | Name | Type | Default | Notes | | ----------------- | ------ | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------- | | `Prompt` | Text | — | Required | | `Reference Image` | Upload | Optional | Used for edits and variations | | `Size` | Select | Square (1:1) | Square (1:1), Landscape (16:9), Landscape (4:3), Landscape (3:2), Portrait (2:3), Portrait (3:4), Portrait (9:16), Ultra-wide (21:9), Tall (9:21) | | `Resolution` | Select | 2K | 2K, 4K | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------- | ----------------------- | -------------------------- | | Text → Image | \~25 seconds | Prompt | High-resolution generation | | Image → Image | \~25 seconds | Reference Image, Prompt | Inpainting & edits | *** ### **Output options** | Option | Values / notes | | ------------ | --------------------------------------------------------------- | | Aspect ratio | Square (1:1), Landscape (16:9), Portrait (9:16) | | Resolution | 2048×2048 (Square), 2048×1152 (Landscape), 1152×2048 (Portrait) | *** ### **Prompt tips** * Use photographic tokens for realism. *** # Seedream 4.5 (by Bytedance) ### **Overview** A new-generation image creation model with enhanced quality and multi-image blending capabilities. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) · Images → Image (multi-image blend) * **Outputs / sizes:** 1024×1024 (1:1), 1280×720 (16:9), 720×1280 (9:16) — higher when selecting 2K/4K * **Aspect ratios:** 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 *** ### **What it's great for** * High-quality photoreal images with improved generation fidelity. * Multi-image blending and composition (up to 10 input images). * Image editing, inpainting, and style transfer tasks. * High-resolution 4K asset creation. *** ### **Parameters (editor-ready)** | Name | Type | Default | Notes | | ----------------- | ------ | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------- | | `Prompt` | Text | — | Required | | `Reference Image` | Upload | Optional | Used for edits and variations (up to 10 images for multi-image mode) | | `Aspect Ratio` | Select | Square (1:1) | Square (1:1), Landscape (16:9), Landscape (4:3), Landscape (3:2), Portrait (2:3), Portrait (3:4), Portrait (9:16), Ultra-wide (21:9), Tall (9:21) | | `Resolution` | Select | 2K | 2K, 4K | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | -------------- | -------------- | ----------------------- | ---------------------------------- | | Text → Image | \~25 seconds | Prompt | High-resolution generation | | Image → Image | \~25 seconds | Reference Image, Prompt | Inpainting & edits | | Images → Image | \~25 seconds | Multiple Images, Prompt | Multi-image blending & composition | *** ### **Output options** | Option | Values / notes | | ------------ | --------------------------------------------------------------- | | Aspect ratio | Square (1:1), Landscape (16:9), Portrait (9:16), and more | | Resolution | 2048×2048 (Square), 2048×1152 (Landscape), 1152×2048 (Portrait) | *** ### **Prompt tips** * Use photographic tokens for realism. * Describe lighting, camera angles, and mood for cinematic results. * For multi-image blending, describe how elements should combine in your prompt. *** # Magnific Upscaler & Enhancer ### **Overview** Image upscaler and enhancer for high-detail restoration and creative upscaling. *** ### **Quick facts** * **Modes:** Image → Image (upscale / enhance / transform) * **Default output / size:** Precision 2× (fixed) · Creative 2×–16× * **Aspect ratios:** Preserves input aspect ratio * **Variants:** Precision (detail-first) · Creative (promptable stylization) *** ### **When to use each model** * **Magnific Precision:** Best when you must preserve the original look—restoring archival scans, print prep, or broadcast stills where only a 2× upscale is needed. Fine-tune sharpness, grain, and ultra-detail without prompting. * **Magnific Creative:** Choose this when you want to reinterpret the source—concept art, stylized photography, marketing assets, or when you need 4×–16× upscales. Add a prompt, pick an optimization preset, and steer creativity, detail, resemblance, and fractality. *** ### **What it’s great for** * Photo upscaling for print and large-format uses. * Enhancing texture, denoise, and creative detail reimagination. * Preparing assets for VFX or large displays. *** ### **Example outputs** {% hint style="info" %} Low-resolution input sample {% endhint %}
{% hint style="info" %} Upscaled result — 4× detail restoration {% endhint %}
#### **Copy-and-paste prompts** ``` Upscale image to 4x, preserve skin detail, remove JPEG artifacts ``` ``` Enhance texture and noise removal for print, keep natural lighting ``` ### **Parameters** #### **Magnific Precision parameters** | Name | Type | Default | Notes | | -------------- | ------ | ------- | ------------------------------------------------------------- | | `Image` | Upload | — | Required input image | | `Scale Factor` | Select | 2× | Fixed 2× enlargement (API enforces max allowed scale) | | `Sharpen` | Slider | 50 | 0–100; higher values accentuate edges but can look artificial | | `Smart Grain` | Slider | 7 | 0–100; reintroduces natural texture for photo realism | | `Ultra Detail` | Slider | 30 | 0–100; boosts fine structure and clarity | #### **Magnific Creative parameters** | Name | Type | Default | Notes | | --------------------- | ------ | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ | | `Prompt` | Text | — | Optional guidance for creative reinterpretation | | `Image` | Upload | — | Required input image | | `Optimization Preset` | Select | Standard | Standard, Soft Portraits, Hard Portraits, Art & Illustration, Video Game Assets, Nature & Landscapes, Films & Photography, 3D Renders, Sci-Fi & Horror | | `Scale Factor` | Select | 2× | 2×, 4× (2× cost), 8× (4× cost), 16× (8× cost); engine auto-limits if resolution would exceed 10056×10056 | | `Creativity` | Slider | 2 | -10 to 10; higher values introduce more generative edits | | `Detail (HDR)` | Slider | 6 | -10 to 10; controls definition and micro-contrast | | `Resemblance` | Slider | 6 | -10 to 10; keeps the output closer to the original when lowered | | `Fractality` | Slider | 6 | -10 to 10; governs prompt strength and intricate patterns | | `Engine` | Select | Auto | Auto, Illusio (illustrations), Sharpy (photos), Sparkle (balanced; reduces JPEG artifacts) | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ----------------------- | -------------- | --------------- | -------------------------- | | Image → Image (upscale) | \~15 seconds | Image | Upscale for print & detail | | Image → Image (enhance) | \~15 seconds | Image | Denoise and texture work | *** ### **Output options** | Option | Values / notes | | ------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Scale | Precision: 2× fixed · Creative: 2×, 4×, 8×, 16× (auto-clamped if output would exceed 10056×10056) | | Formats | PNG, JPG | | Notes | Creative presets: Standard, Soft Portraits, Hard Portraits, Art & Illustration, Video Game Assets, Nature & Landscapes, Films & Photography, 3D Renders, Sci-Fi & Horror | *** ### **Prompt tips** * For Precision, start with the base 2× upscale and nudge Sharpen/Ultra Detail carefully—small increments keep it photoreal. * For Creative, pair a succinct prompt with an Optimization preset that matches the content (e.g., Soft Portraits for faces). * Dial Resemblance down if you want a bolder reimagining; push Fractality up only when you need intricate stylized detail. # Topaz ## **Topaz (Photo & Video enhancement)** **Professional photo & video enhancement tools for de-noise, de-blur, and upscaling.** *** ### **Quick facts** * **Modes:** Image → Image (upscale / denoise / deblur) · Video → Video (upscaling) * **Default output / size:** Image 2× (range 1×–4×) · Video 2× (range 1×–4×) * **Aspect ratios:** Preserves input aspect ratio (image & video) * **Enhancement presets:** Standard V2, Low Resolution V2, CGI, High Fidelity V2, Text Refine *** ### **When to use each mode** * **Topaz Photo (image):** Best for photographers and designers who need artifact-free enlargements, text sharpening, or restoration without touching motion settings. * **Topaz Video:** Use for documentary, broadcast, or archival clips that need 2×–4× upscale or frame-rate adjustments while keeping motion smooth. *** ### **What it’s great for** * Photographers and studios preparing images for print. * Restoring and cleaning archival photos/video. * High-quality video upscaling and restoration. *** #### **Topaz Tips** ``` Upscale to 2x, remove noise while preserving textures ``` ``` Restore scanned photo: remove dust and scratches, sharpen faces ``` ``` Video upscale: 1080p -> 4K, maintain motion clarity ``` *** ### **Parameters (editor-ready)** #### **Topaz Photo (image)** | Name | Type | Default | Notes | | ------------- | ------ | -------- | ---------------------------------------------------------------------------- | | `Image` | Upload | — | Required still image input | | `Enhancement` | Select | Standard | Standard V2 (default), Low Resolution V2, CGI, High Fidelity V2, Text Refine | | `Scale` | Slider | 2× | 1×–4× (0.5 increments); higher values increase cost and render time | | `Seed` | Seed | Random | Optional; set for reproducible enhancement | #### **Topaz Video** | Name | Type | Default | Notes | | -------------- | --------- | ------- | ----------------------------------------------------- | | `Video` | Upload | — | Input clip (≤8192×8192) | | `Scale` | Slider | 2× | 1×–4× (0.5 increments); requests above 4× are clamped | | `Target FPS` | Slider | 24 | 16–60 FPS; retimes output to match selection | | `Aspect Ratio` | Inherited | Auto | Hidden control; keeps the original frame proportions | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------- | --------------- | ------------------------------- | | Image → Image | \~10 seconds | Media | Upscale & restore photos | | Video → Video | \~1 minute | Media | Video upscaling & stabilization | *** ### **Output options** | Option | Values / notes | | ------- | --------------------------------------------------------------------------------------- | | Scale | Image: 1×–4× · Video: 1×–4× (auto-clamped to max 8192×8192 output) | | Formats | PNG, JPG (image) · MP4 (video) | | Notes | Enhancement presets: Standard V2, Low Resolution V2, CGI, High Fidelity V2, Text Refine | *** ### **Prompt tips** * For portraits or text, start with Standard V2; switch to Low Resolution V2 for tiny originals or Text Refine for signage. * Preview a short crop before long renders to dial in scale, FPS, and enhancement choices. *** # Flux Kontext Max (by Black Forest Labs) ### **Overview** High performing text-to-image model with multi-image support and flexible aspect ratios. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image · Images → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 *** ### **What it's great for** * High-quality text-to-image generation with excellent prompt adherence * Multi-image blending and composition tasks * Image editing with reference images * Creative workflows requiring diverse aspect ratios *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | ----------------------------------------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 (auto for image inputs) | | `Seed` | Number | Random | Optional for repeatable results | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | -------------- | -------------: | --------------- | ------------------------- | | Text → Image | \~8s | Prompt | Generate images from text | | Image → Image | \~8s | Prompt, Image | Edit or transform images | | Images → Image | \~8s | Prompt, Images | Blend up to 5 images | *** ### **Output options** | Option | Values / notes | | ------------ | ----------------------------------------------- | | Aspect ratio | 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 | *** ### **Prompt tips** * Use specific style descriptors for better results * Seed values enable reproducible outputs * For multi-image blending, describe the desired composition clearly *** ### **Safety** Text moderation is applied according to platform rules — avoid disallowed content. *** # Flux Pro (by Black Forest Labs) ### **Overview** High performing text-to-image model with exceptional quality and detail. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 *** ### **What it's great for** * Professional-grade image generation with superior quality * Complex prompts requiring high adherence and detail * Commercial and creative projects demanding best-in-class results * Wide range of artistic and photographic styles *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | ----------------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 | | `Seed` | Number | Random | Optional for repeatable results | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ------------------------------------- | | Text → Image | \~24s | Prompt | Professional text-to-image generation | *** ### **Output options** | Option | Values / notes | | ------------ | ----------------------------------------------- | | Aspect ratio | 21:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:21 | *** ### **Prompt tips** * Detailed prompts produce better results with this model * Style descriptors work exceptionally well * Seed values enable exact reproduction * Longer generation time reflects higher quality processing *** ### **Safety** Text moderation is applied according to platform rules — avoid disallowed content. *** # Flux Redux (by Black Forest Labs) ### **Overview** Image variation model for slight modifications of input images. *** ### **Quick facts** * **Modes:** Image → Image * **Default output / size:** Auto (preserves input aspect) * **Aspect ratios:** Auto + multiple resolution options *** ### **What it's great for** * Creating subtle variations of existing images * Iterative refinement workflows * Maintaining overall composition while varying details * Quick image alternatives *** ### Parameters | Control | Type | Default | Notes | | ----------------- | ------ | ------- | ------------------------------- | | `Reference Image` | Upload | — | Input image to vary (required) | | `Quality` | Slider | 100 | 0-100 | | `Size` | Select | Auto | Auto, 16:9, 4:3, 1:1, 3:4, 9:16 | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | --------------- | | Image → Image | \~28s | Image | Image variation | *** ### **Output options** | Option | Values / notes | | ------------ | ------------------------------------------------------ | | Aspect ratio | Auto (preserves input) + multiple aspect ratio options | *** ### **Prompt tips** * No prompt required—model creates variations automatically * Lower guidance scale produces more variation * Higher guidance scale stays closer to input * Useful for generating alternatives quickly *** ### **Safety** Moderation may be applied according to platform rules — avoid disallowed content. *** # Ideogram 3.0 ### **Overview** Stylized, customizable text-to-image model with extensive preset options. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** 3:1, 16:10, 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16, 10:16, 1:3 *** ### **What it's great for** * Stylized image generation with curated aesthetic presets * Creative projects requiring distinctive visual styles * Text-in-image rendering capabilities * Flexible aspect ratios for different use cases *** ### Parameters | Control | Type | Default | Notes | | ---------------- | -------- | ------- | ----------------------------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | 3:1, 16:10, 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16, 10:16, 1:3 | | `Ideogram Style` | Dropdown | None | Multiple curated style options available | | `Seed` | Number | Random | Optional for repeatable results | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ------------------------- | | Text → Image | \~18s | Prompt | Stylized image generation | *** ### **Output options** | Option | Values / notes | | ------------ | ----------------------------------------------------------- | | Aspect ratio | 3:1, 16:10, 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16, 10:16, 1:3 | *** ### **Prompt tips** * Explore different style presets for varied aesthetic results * Model excels at rendering text within images * Wide aspect ratio support enables diverse compositions *** ### **Safety** Text moderation is applied according to platform rules — avoid disallowed content. *** # Ideogram Character ### **Overview** Character consistency and face swapping with reference images. *** ### **Quick facts** * **Modes:** Image → Image · Images → Image * **Default output / size:** 1024×1024 (Auto/1:1) * **Aspect ratios:** Auto, 3:1, 16:10, 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16, 10:16, 1:3 *** ### **What it's great for** * Maintaining character consistency across generations * Face swapping with reference images * Character-driven creative projects * Consistent character appearances in different scenes *** ### Parameters | Control | Type | Default | Notes | | -------------------- | -------- | ------- | ----------------------------------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Reference Image(s)` | Upload | — | 1-2 images (1 for editing, 2 for face swap) | | `Aspect Ratio` | Dropdown | Auto | Auto, 3:1, 16:10, 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16, 10:16, 1:3 | | `Mode` | Dropdown | Auto | Auto, Fiction, Realistic | | `Seed` | Number | Random | Optional for repeatable results | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | -------------- | -------------: | ---------------- | --------------------- | | Image → Image | \~18s | Prompt, Image | Character consistency | | Images → Image | \~18s | Prompt, 2 Images | Face swapping | *** ### **Output options** | Option | Values / notes | | ------------ | ----------------------------------------------------------------- | | Aspect ratio | Auto, 3:1, 16:10, 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16, 10:16, 1:3 | *** ### **Prompt tips** * Use reference images to maintain character consistency * For face swapping, provide input image first, character reference second * Style Type affects realism vs stylization *** ### **Safety** Text moderation is applied according to platform rules — avoid disallowed content. *** # Luma Photon ### **Overview** Diverse text-to-image model with good prompt adherence. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple aspect ratios supported *** ### **What it's great for** * Versatile image generation across diverse styles * Strong prompt adherence and interpretation * Fast generation times * Creative and commercial projects *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | — | Multiple options available | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ------------------------ | | Text → Image | \~20s | Prompt | Diverse image generation | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Model handles a wide range of styles effectively * Clear descriptors improve prompt adherence * Good balance between creativity and accuracy * Fast iteration times for experimentation *** ### **Safety** Moderation may be applied according to platform rules — avoid disallowed content. *** # GPT Image 1.5 (by OpenAI) ### **Overview** High-fidelity image generation with strong prompt adherence and transparent background support. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** 3:2, 1:1, 2:3 *** ### **What it's great for** * Premium quality image generation with precise control. * Images with transparent backgrounds for design assets. * Strong prompt adherence for detailed specifications. * High-fidelity outputs for professional use cases. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | 3:2, 1:1, 2:3 | | `Quality` | Dropdown | High | Low, Medium, High | | `Background` | Dropdown | Auto | Auto, Transparent, Opaque | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ---------------------------- | | Text → Image | \~30s | Prompt | High-fidelity generation | | Image → Image | \~35s | Image, Prompt | Editing with strong fidelity | *** ### **Output options** | Option | Values / notes | | ------------ | ------------------------- | | Aspect ratio | 3:2, 1:1, 2:3 | | Background | Auto, Transparent, Opaque | *** ### **Prompt tips** * Specify composition, lighting, and desired ratio for best results. * Use "Transparent" background for design assets and icons. * Model excels at following detailed, specific prompts. *** # Imagen 4 (by Google) ### **Overview** Advanced image generation model by Google DeepMind with high-quality outputs. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** 1:1, 3:4, 4:3, 9:16, 16:9 *** ### **What it's great for** * High-quality image generation with Google's latest AI. * Photorealistic outputs with strong detail. * Versatile aspect ratios for different use cases. * Professional-grade image creation. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | 1:1, 3:4, 4:3, 9:16, 16:9 | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ----------------------- | | Text → Image | \~20s | Prompt | High-quality generation | *** ### **Output options** | Option | Values / notes | | ------------ | ------------------------- | | Aspect ratio | 1:1, 3:4, 4:3, 9:16, 16:9 | *** ### **Prompt tips** * Descriptive prompts yield best results. * Model handles both photorealistic and artistic styles. * Specify aspect ratio based on intended use. *** ### **Safety** Text moderation is applied according to Google's usage policies. *** # Wan 2.2 (by Alibaba) ### **Overview** High realism image model that is coherent, controllable, and cost-effective. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Cost-effective realistic image generation. * Coherent outputs with good controllability. * Budget-friendly option for high-volume generation. * Realistic imagery across various subjects. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | -------------------- | | Text → Image | \~12s | Prompt | Realistic generation | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Model handles realistic subjects well. * Clear, descriptive prompts yield best results. * Cost-effective for iterative workflows. *** ### **Safety** Moderation is applied according to platform rules. *** # Recraft V3 ### **Overview** Leading model for design work with high consistency and creative flexibility. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Concept art and creative storytelling. * Design-focused image generation. * Highly consistent outputs across generations. * Professional creative and commercial projects. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | ----------------- | | Text → Image | \~14s | Prompt | Design generation | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Model excels at design and concept art styles. * Consistent results make it ideal for series work. * Clear style direction improves outputs. *** # Stable Diffusion 3.5 ### **Overview** Versatile image model offering large output variety and rapid iteration. *** ### **Quick facts** * **Modes:** Text → Image * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * General-purpose image generation. * Rapid iteration and experimentation. * Wide variety of styles and subjects. * Flexible creative exploration. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------ | -------------: | --------------- | -------------------------- | | Text → Image | \~32s | Prompt | General-purpose generation | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Model handles a wide range of styles effectively. * Good for exploration and finding creative directions. * Supports various artistic and photorealistic styles. *** # Nano Banana Pro (by Google) ### **Overview** Google's most powerful image model with premium 4K generation and multi-image composition capabilities. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) · Images → Image (multi-image blend) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** 21:9, 16:9, 5:4, 4:3, 3:2, 1:1, 2:3, 3:4, 4:5, 9:16 * **Resolutions:** 1K, 2K, 4K *** ### **What it's great for** * Premium quality image generation at up to 4K resolution. * Professional multi-image composition (up to 14 input images). * Highest fidelity outputs for commercial and professional use. * Complex editing and inpainting tasks. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | 21:9, 16:9, 5:4, 4:3, 3:2, 1:1, 2:3, 3:4, 4:5, 9:16 | | `Resolution` | Dropdown | 1K | 1K, 2K, 4K (4K costs 2x) | | `Image` | Upload | — | Required for editing modes | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | -------------- | -------------: | ----------------------- | -------------------------- | | Text → Image | \~30s | Prompt | Premium quality generation | | Image → Image | \~35s | Image, Prompt | High-fidelity editing | | Images → Image | \~40s | Multiple Images, Prompt | Multi-image composition | *** ### **Output options** | Option | Values / notes | | ------------ | --------------------------------------------------- | | Aspect ratio | 21:9, 16:9, 5:4, 4:3, 3:2, 1:1, 2:3, 3:4, 4:5, 9:16 | | Resolution | 1K, 2K, 4K | *** ### **Prompt tips** * Model excels at photorealistic and artistic outputs. * Use 4K resolution for maximum detail in final assets. * Multi-image mode supports up to 14 input images for complex compositions. *** ### **Safety** Text moderation is applied according to Google's usage policies. *** # Z-Image Turbo (by Alibaba) ### **Overview** Super-fast, cost-effective image model for rapid iteration and prototyping. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Rapid iteration and quick prototyping. * Budget-friendly image generation at scale. * Fast turnaround for draft concepts. * High-volume generation workflows. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | | `Image` | Upload | — | Required for editing mode | | `Strength` | Slider | 0.8 | Controls edit intensity (I2I mode) | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ---------------- | | Text → Image | \~4s | Prompt | Rapid generation | | Image → Image | \~5s | Image, Prompt | Quick edits | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Great for quickly testing concepts before using premium models. * Ideal for high-volume workflows where speed matters. * Supports automatic prompt translation to English. *** # Kling O1 ### **Overview** Precise image editing model for detailed modifications and refinements. *** ### **Quick facts** * **Modes:** Image → Image (editing) * **Default output / size:** Matches input image * **Aspect ratios:** Auto (matches input) *** ### **What it's great for** * Precise image edits with high accuracy. * Detailed modifications and refinements. * Professional retouching workflows. * Targeted adjustments to specific areas. *** ### Parameters | Control | Type | Default | Notes | | -------- | ------ | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Image` | Upload | — | Required - the image to edit | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | --------------- | | Image → Image | \~20s | Image, Prompt | Precise editing | *** ### **Output options** | Option | Values / notes | | ------------ | -------------- | | Aspect ratio | Matches input | *** ### **Prompt tips** * Be specific about the changes you want to make. * Model excels at targeted, precise modifications. * Works well for professional editing workflows. *** # Qwen Image Edit Plus ### **Overview** Superior text editing and image composition model with advanced capabilities. *** ### **Quick facts** * **Modes:** Image → Image (editing) · Images → Image (multi-image) * **Default output / size:** Matches input or 1024×1024 * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Complex text editing within images. * Advanced image composition and blending. * Multi-image workflows requiring precision. * Professional editing with superior quality. *** ### Parameters | Control | Type | Default | Notes | | -------- | ------ | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Image` | Upload | — | Required - the image(s) to edit | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | -------------- | -------------: | ----------------------- | ------------------------ | | Image → Image | \~26s | Image, Prompt | Text & composition edits | | Images → Image | \~30s | Multiple Images, Prompt | Multi-image composition | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Model excels at editing text within images. * Supports complex multi-image compositions. * Use clear, specific instructions for best results. *** # Qwen Image Edit Angles ### **Overview** Edit images from different angles and perspectives with advanced viewpoint control. *** ### **Quick facts** * **Modes:** Image → Image (editing) * **Default output / size:** Matches input * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Multi-angle image editing. * Perspective adjustments and transformations. * Creating alternative viewpoints from existing images. * Professional workflows requiring angle variations. *** ### Parameters | Control | Type | Default | Notes | | -------- | ------ | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Image` | Upload | — | Required - the image to edit | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------- | | Image → Image | \~28s | Image, Prompt | Angle-based editing | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Describe the desired angle or perspective change clearly. * Model understands spatial transformations well. * Works best with clear subjects and defined compositions. *** # Flux 2 (by Black Forest Labs) ### **Overview** Updated BFL model with high customizability and acceleration options. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Fast, customizable image generation. * Acceleration options for speed optimization. * Cost-effective high-quality outputs. * Both generation and editing workflows. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | | `Acceleration` | Dropdown | Regular | None, Regular, High | | `Quality` | Slider | 40 | Inference steps (10-40) | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | --------------- | | Text → Image | \~8s | Prompt | Fast generation | | Image → Image | \~10s | Image, Prompt | Quick editing | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Use acceleration options to optimize for speed vs quality. * Model handles a wide range of styles effectively. * Good balance of speed and quality for iterative workflows. *** # Flux 2 Max (by Black Forest Labs) ### **Overview** Premium Flux 2 variant with exceptional realism, precision, and consistency. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Premium photoreal outputs with maximum quality. * Exceptional consistency across generations. * Professional-grade image creation. * High-fidelity commercial work. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | | `Acceleration` | Dropdown | Regular | None, Regular, High | | `Quality` | Slider | 40 | Inference steps (10-40) | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | -------------------- | | Text → Image | \~12s | Prompt | Premium generation | | Image → Image | \~15s | Image, Prompt | High-quality editing | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Best for final assets requiring maximum quality. * Exceptional at photorealistic outputs. * Use for commercial and professional projects. *** # Flux 2 Flex (by Black Forest Labs) ### **Overview** Highly consistent Flux 2 variant optimized for reliable, reproducible outputs. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Consistent outputs across multiple generations. * Reliable results for batch workflows. * Professional projects requiring reproducibility. * Series work with consistent style. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | | `Quality` | Slider | 40 | Inference steps (10-40) | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | --------------------- | | Text → Image | \~10s | Prompt | Consistent generation | | Image → Image | \~12s | Image, Prompt | Reliable editing | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Ideal for creating series with consistent style. * Great for batch workflows requiring reproducibility. * Use for projects where consistency is critical. *** # Flux 2 Pro (by Black Forest Labs) ### **Overview** Highest performing model from Black Forest Labs with professional-grade quality. *** ### **Quick facts** * **Modes:** Text → Image · Image → Image (editing) * **Default output / size:** 1024×1024 (1:1) * **Aspect ratios:** Multiple standard ratios supported *** ### **What it's great for** * Professional-grade image generation. * Best overall quality from BFL. * Commercial and creative projects. * High-fidelity outputs for final assets. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Aspect Ratio` | Dropdown | 1:1 | Multiple options available | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ----------------------- | | Text → Image | \~10s | Prompt | Professional generation | | Image → Image | \~12s | Image, Prompt | High-quality editing | *** ### **Output options** | Option | Values / notes | | ------------ | -------------------------------- | | Aspect ratio | Multiple aspect ratios supported | *** ### **Prompt tips** * Best overall quality from the Flux 2 family. * Ideal for final production assets. * Handles both creative and photorealistic styles well. *** # Imagen 3 (by Google) ### **Overview** High-quality image editing model by Google DeepMind. *** ### **Quick facts** * **Modes:** Image → Image (editing) * **Default output / size:** Matches input or 1024×1024 * **Aspect ratios:** Auto, 1:1, 3:4, 4:3, 9:16, 16:9 *** ### **What it's great for** * High-quality image editing powered by Google. * Professional editing and retouching. * Style transfer and image modifications. * Precise edits with natural results. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | --------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Image` | Upload | — | Required - the image to edit | | `Aspect Ratio` | Dropdown | Auto | Auto, 1:1, 3:4, 4:3, 9:16, 16:9 | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------ | | Image → Image | \~20s | Image, Prompt | High-quality edits | *** ### **Output options** | Option | Values / notes | | ------------ | ------------------------------- | | Aspect ratio | Auto, 1:1, 3:4, 4:3, 9:16, 16:9 | *** ### **Prompt tips** * Describe the desired edit clearly. * Model excels at natural-looking modifications. * Works well for both subtle and significant changes. *** ### **Safety** Text and image moderation applied according to Google's usage policies. *** # Video Models Here are all the video models currently available in FLORA. ## Generation Models These models generate video from text or image inputs.
ModelEst. time (s)DescriptionBest for
Aurora30Animate an avatar using an image and audio input.Animate an avatar using an image and audio input.
Frame Morphing210Creative morph between two images.Smooth transitions, creative blending between images
Gemini-Omni-Flash70Google's state of the art, Omni video model.Google's state of the art, Omni video model.
Grok Imagine115Video generation with audio from xAI.Quick generation with xAI's latest image capabilities
Happy Horse 1.0250Generate videos with synced audio using customizable parameters.Generate videos with synced audio using customizable parameters.
Kling 2.0 Master330Reliable motion with good consistency.Complex long-form sequences, cinematic pre-viz, style explorations.
Kling 2.1 Master330Precise prompt following with dynamic motion.TikTok-style transitions, before-and-after promos, precise framing.
Kling 2.1 Pro330Precise prompt following with dynamic motion.High-quality generation with strong prompt adherence.
Kling 2.5 Turbo Pro250Smooth motion at faster generation speeds.Fast, high-quality generation with smooth motion.
Kling 2.6 Pro200Top-tier visuals with synced voice and sound.High-quality videos with synchronized audio, music videos, commercials.
Kling 3.0 Pro270Animate between start and end frames.Premium video with synchronized audio, cinematic quality.
Kling 3.0 Pro (Turbo)270Improved lipsync and multishot generation capabilities.Improved lipsync and multishot generation capabilities.
Kling 3.0 Standard220Faster and more cost effective than Kling 3.0 Pro.Quick high-quality generation at lower cost.
Kling 3.0 Standard (Turbo)270Optimized for rapid iteration and high-volume production.Optimized for rapid iteration and high-volume production.
Kling Avatar v2 Pro300Create avatar videos with realistic humans, animals, cartoons, or stylized characters.Avatar-driven storytelling, character videos, and stylized spokesperson content.
Kling O1330Precise edits with text, image, or video guidance.Detailed image editing with high precision
Kling O3 Pro265Animate between start and end frames.Maximum quality Kling output, premium productions.
Kling O3 Standard220Kling O3 at a lower cost and faster speed.Cost-effective O3 quality, faster generation.
Kling Pro 1.5300Strong motion with dynamic camera work.Dynamic camera movements, action sequences.
Kling Pro 1.6360Balanced quality and generation speed.Product demos, architectural walk-throughs, realistic lighting.
LTX-2 Pro130Fast 4K output with generated audio.Quick video with synchronized audio, up to 4K resolution.
Luma Ray 2200Realistic physics and natural motion.Lifestyle, nature, travel footage, drone-style glides, fashion runways.
Luma Ray 2 Flash603× faster than Ray 2, good quality.Quick variant testing, social-media cut-downs, preview edits.
Marey420Commercially-licensed training data.Brand-safe commercial spots, legal-clear hero sequences.
Minimax Hailuo240Strong motion with dynamic camera work.Narrative shorts (30-60s), coherent character arcs, evolving lighting.
Minimax Hailuo 02 Pro420Native 1080p with realistic physics.Ultrafast generation of vertical reels, bumper ads, A/B variants.
Minimax Hailuo 2.3 Pro3001080p native with expressive motion.High-resolution dynamic content, expressive animations.
Pika350Reliable prompt following, up to 10s clips.Avatar-led explainers, talking-head shorts, expressive faces.
Runway Gen-4.5180High-fidelity cinematic motion with strong prompt adherence.Cinematic output, consistent character and scene quality.
Seedance 1.0 Pro165High quality with fixed camera option.Premium social ads, stylised title sequences, cinematic game-trailers.
Seedance 1.5 Pro120Powerful model with native sound.High quality with built-in audio generation.
Seedance 2.0250Generate a video by setting its start and end frame, via Enhancor.Premium social ads, cinematic sequences with synchronized audio.
Seedance 2.0 Fast175Generate a video by setting its start and end frame, via Enhancor.Quick iterations on Seedance 2.0 quality content.
Seedance 2.0 Mini250Lightweight, cost-effective video generation with audio.Lightweight, cost-effective video generation with audio.
Seedance 2.5300Single-shot video up to 30 seconds with native audio.Single-shot video up to 30 seconds with native audio.
Sora 2 Pro830OpenAI's best, with synced audio and voice.Premium AI video generation, cinematic quality, up to 1080p resolution.
Tencent Hunyuan240Open-source model, customizable for fine-tuning.Multi-locale campaigns, culturally nuanced content, synced soundtracks.
VEED Subtitles30Render subtitles onto videos with selectable style preset and source language.Render subtitles onto videos with selectable style preset and source language.
Veo 2180Good balance of quality and speed.Hero shots, trailers, architecture fly-throughs, luxury product reveals.
Veo 3250Native audio with cinematic quality.4K trailer sequences, high-end commercials, or narrative teasers.
Veo 3.1305Google's best: audio, 1080p, up to 8s clips.Premium productions, cinematic sequences, highest-quality outputs for professional use.
Veo 3.1 (Fast)360Veo 3.1 quality at lower cost.Quick iterations on premium content, cost-effective previews before final renders.
Veo 3.1 Frames250Animate between start and end frames.Precise keyframe animation, controlled transitions.
Veo 3.1 Frames (Fast)390Fast frame-to-frame animation.Quick keyframe previews, rapid iteration on transitions.
Veo 3.1 Frames Lite135More cost effective frame-to-frame animation.Budget-friendly keyframe animation.
Veo 3.1 Ingredients250Combine reference images into video.Multi-reference video generation, consistent character and style.
Veo 3.1 Lite120Veo 3.1 with more cost effective generations.Budget-friendly premium video, scaled production workflows.
WAN 2.2120Diverse motion styles and camera options.Cost-effective realistic image generation
WAN 2.5250Built-in audio generation.High-quality generation, improved motion, versatile styles.
WAN 2.6250Uses audio input to drive the video generation.Complex sequences, multi-shot videos, high-resolution outputs.
WAN 2.7250Uses audio input to drive the video generation.Cinematic multi-shot storytelling with built-in audio generation.
## Editing & Reference Models These models edit existing videos or use reference inputs for guided generation.
ModelEst. time (s)DescriptionBest for
Animatediff300Reanimate video with new motion.Rapid motion-graphics, looping GIFs, meme-style clips.
Fabric 1.060Flagship lipsync model using image and audio inputs.Flagship lipsync model using image and audio inputs.
Grok Imagine Edit120Edit a video using text prompts.Quick edits powered by xAI
Grok Imagine References85Generate a video guided by one or more reference images.Reference-guided video generation with xAI.
Happy Horse 1.0 Edit250Generate videos with synced audio using customizable parameters.Generate videos with synced audio using customizable parameters.
Happy Horse 1.0 References290Use up to 9 images as references to generate a video with integrated audio.Use up to 9 images as references to generate a video with integrated audio.
Kling 2.6 Pro Motion Control96Animate character from reference.Character animation with 2.6 Pro quality.
Kling 3.0 Pro Motion Control655Animate character from reference.Character animation with precise motion control.
Kling 3.0 Standard Motion Control560Animate character from reference.Cost-effective character animation.
Kling O1 Edit450Edit video with text or image guidance.Precise video editing with multimodal guidance.
Kling O1 Reference450Style transfer from a reference video.Video style transfer, consistent aesthetic application.
Kling O1 References255Keep subjects consistent across scenes.Multi-reference consistency, character continuity.
Kling O3 Pro Edit450Flagship video editing from Kling.Premium video editing with maximum quality.
Kling O3 Pro Reference450Use an input video and images as reference.Reference-guided generation with highest quality.
Kling O3 Pro References385Keep subjects consistent across scenes.Multi-reference consistency at premium quality.
Kling O3 Standard Edit450Kling O3 at a lower cost and faster speed.Cost-effective video editing with O3 capabilities.
Kling O3 Standard Reference450Kling O3 at a lower cost and faster speed.Budget-friendly reference-guided generation.
Kling O3 Standard References265Kling O3 at a lower cost and faster speed.Cost-effective multi-reference consistency.
Lipsync 2 Pro300Lipsync an input video performance with input speech.Lip synchronization, dubbed content, talking-head alignment.
LTX-2 Retake50Retake segments of an existing LTXV 2 video.Segment-level video editing and regeneration.
Lucy Edit Pro120Text-prompted video editing.Quick text-driven video modifications and adjustments.
Luma Modify Video205Restyle with adhere, flex, or reimagine modes.Day-for-night, illustrative repainting, live-action to anime transformations.
Marey Motion Transfer400Apply motion from one video to another.Choreography transfer, steadicam paths to new scenes.
Marey Pose Transfer300Copy body poses to new subjects.Digital-double pose matching, stunt-viz, freeze-frame-to-motion.
Seedance 2.0 Reference390Generate video guided by reference imagesReference-guided video generation with audio.
Seedance 2.0 Reference (Fast)235Generate video guided by a single reference image, fast model via Enhancor.Quick reference-guided video iteration.
Sync 3300Next-gen lipsync model with studio-grade capabilities.Next-gen lipsync model with studio-grade capabilities.
VEED Lipsync60Generate realistic lipsync using an input audio file and a video.Generate realistic lipsync using an input audio file and a video.
WAN 2.2 Animate Move1535Apply motion to reference character.Character animation from reference, motion transfer.
WAN 2.2 Animate Replace1440Swap subjects using reference image.Subject replacement in video, face and character swaps.
## Upscaling & Post-Processing These models enhance, upscale, or post-process existing videos.
ModelEst. time (s)DescriptionBest for
Bria Video Upscaler15Enterprise-safe, commercially licensed.Commercial video upscaling with licensing compliance.
Magnific Creative Turbo Video Upscaler300Upscale videos faster while maintaining visual quality.Fast video upscaling with creative enhancement.
Magnific Creative Video Upscaler490Upscale videos up to 4k resolution.Artistic video upscaling to 4K.
Magnific Precision Video Upscaler490High-accuracy video upscaling.Detail-preserving video upscaling.
Mirelo SFX 1.530Add synced sounds to a video input.Add synced sounds to a video input.
MMaudio v230Add sound to your video.Automatic audio generation from video content.
Topaz Upscaler165Professional video upscaling to 4K.Industry-standard upscaling for professional workflows
VEED Background Removal60Remove the background from your video.Green-screen effects, subject isolation, compositing.
# Veo 3.1 (by Google) ### Overview The most advanced AI video model from Google DeepMind. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Resolutions:** 720p, 1080p * **Aspect ratios:** 16:9, 1:1, 9:16 *** ### What it's great for * Premium productions requiring highest quality. * Cinematic sequences and professional content. * Commercial and enterprise video projects. * Maximum fidelity outputs. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | -------------------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | 16:9, 1:1, 9:16 | | Duration | Dropdown | 8s | Multiple options available | | Resolution | Dropdown | 720p | 720p, 1080p | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------------ | | Text → Video | \~180s | Prompt | Premium video generation | | Image → Video | \~180s | Image, Prompt | Animated sequences | *** ### Prompt tips * Detailed prompts yield best results. * Specify camera movements and lighting for cinematic quality. * Best for final production assets. *** ### Safety Text moderation applied according to Google's usage policies. *** # Veo 3 (by Google) ### **Overview** Advanced text→video model with native audio generation and high-quality outputs. *** ### **Quick facts** * **Modes:** Text → Video · Image → Video. * **Default clip length:** 8 seconds * **Resolution:** 960×960 (square) · 1280×720 (landscape) · 720×1280 (portrait) * **Aspect ratios:** 16:9, 1:1, 9:16 *** ### **What it’s great for** * Short cinematic clips with integrated audio. * Rapid prototyping with sound design included. * Short-form content pipelines. *** ### **Example outputs** {% hint style="info" %} Scene with native audio — fox in misty forest {% endhint %} {% embed url="" %} {% hint style="info" %} Motion frame — high-contrast macro shot {% endhint %} {% embed url="" %} {% hint style="info" %} Audio-synced shot — dialogue moment {% endhint %} {% embed url="" %} #### **Copy-and-paste prompts** ``` "A close-up, cinematic shot of a lone fox in a misty forest, fur rippling with each breath, slow dolly-in as it turns its head toward the camera, faint rustle of leaves and a distant low call filling the soundscape." ``` ``` "High-contrast black and white macro video of an Asian female eye veiled by long strands of hair, shot on a 100mm macro lens with handheld-to-dolly motion; slow 8–10s dolly-in with parallax through hair, subtle rack focuses, micro-blink, moisture beads, drifting dust motes, film-grain,2000s cinema. Ominous dreamcore/mood with single soft key light, deep shadows, crushed blacks, specular pupil gleam, and faint lens/film defects." ``` ``` "A cinematic video of a woman on a dimly lit train station platform, camera dollying in as she whispers, “I’m leaving tonight, and you won’t find me here again,” her voice echoing under flickering lights, blending with footsteps and the rising rumble of an approaching train." ``` *** ### **Parameters** | Name | Type | Default | Notes | | -------------- | ------ | ---------------- | ----------------------------------------------- | | `Prompt` | Text | — | Required | | `Aspect Ratio` | Select | Landscape (16:9) | Landscape (16:9), Square (1:1), Portrait (9:16) | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------- | --------------- | ------------------------- | | Text → Video | \~4 minutes | Prompt | Videos with audio | | Image → Video | \~4 minutes | Image, Prompt | Animate stills with sound | *** ### **Output options** | Option | Values / notes | | ------- | ----------------- | | Formats | MP4 (H.264 + AAC) | *** ### **Prompt tips** * Describe desired soundscapes in the prompt for richer audio results. * Use explicit timing tokens for beats or cuts. *** # Sora 2 Pro (by OpenAI) ### Overview OpenAI's cutting-edge video generation model. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Resolutions:** 720p, 1080p * **Aspect ratios:** 16:9, 9:16 *** ### What it's great for * Premium AI video generation. * Cinematic quality outputs. * Professional content creation. * High-fidelity video production. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | -------------------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | 16:9, 9:16 | | Duration | Dropdown | — | Multiple options available | | Resolution | Dropdown | 720p | 720p, 1080p (1.66x cost) | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------------ | | Text → Video | \~120s | Prompt | Premium video generation | | Image → Video | \~120s | Image, Prompt | Animated sequences | *** ### Prompt tips * OpenAI's most advanced video model. * Excellent for cinematic and narrative content. * Supports detailed scene descriptions. *** # Kling 2.6 Pro (by Kuaishou) ### Overview Top-tier visuals with native sound generation. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Audio:** Native sound generation included * **Aspect ratios:** 16:9, 9:16, 1:1 *** ### What it's great for * High-quality videos with synchronized audio. * Music videos and commercials. * Content requiring natural sound design. * Professional productions with audio. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | ----------------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | 16:9, 9:16, 1:1 | | Duration | Dropdown | — | Multiple options | | Generate Audio | Toggle | On | Native sound generation | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------- | | Text → Video | \~180s | Prompt | Video with audio | | Image → Video | \~180s | Image, Prompt | Animated with sound | *** ### Prompt tips * Model generates synchronized audio automatically. * Describe both visual and audio elements in prompts. * Best for content where sound matters. *** # Kling 2.5 Turbo Pro (by Kuaishou) ### **Overview** Exceptional motion fluidity and visual quality. *** ### **Quick facts** * **Modes:** Text → Video · Image → Video * **Default clip length / output size:** Variable duration options * **Aspect ratios:** 16:9, 9:16, 1:1 *** ### **What it's great for** * Exceptional motion fluidity * High visual quality * Multiple duration options * Text and image-based video generation *** ### **Parameters** | Control | Type | Default | Notes | | ----------------- | -------- | ------- | ---------------------------------------- | | `Prompt` | Text | — | Natural-language instruction (required) | | `Reference Image` | Upload | — | Optional (needed for Image → Video mode) | | `Duration` | Dropdown | — | 5s, 10s | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ---------------------- | | Text → Video | \~200s | Prompt | Fluid motion from text | | Image → Video | \~150s | Prompt, Image | Animate images | *** ### **Output options** | Option | Values / notes | | -------- | -------------- | | Duration | 5s, 10s | *** ### **Prompt tips** * Model excels at fluid motion sequences * Clear motion descriptions produce best results * Works well with dynamic scenes *** ### **Safety** Moderation may be applied according to platform rules — avoid disallowed content. *** # Kling 2.1 Master (by Kuaishou) ### **Overview** Text → video model for motion-rich short clips. *** ### **Quick facts** * **Modes:** Text → Video · Image → Video * **Default clip length:** 5 seconds (optionally 10s) * **Aspect ratios:** 9:16 *** ### **What it’s great for** * Short cinematic motion generation. * Motion diversity and prompt adherence in quick clips. * Iterative creative experiments across quality tiers. *** ### **Example outputs** {% hint style="info" %} Image reference — deer close-up {% endhint %}
{% hint style="info" %} Image reference video {% endhint %} {% embed url="" %} {% hint style="info" %} 5s text → video motion shot {% endhint %} {% embed url="" %} {% hint style="info" %} Action motion frame {% endhint %} {% embed url="" %} #### **Copy-and-paste prompts** ``` "Fashion editorial close-up video: lower body mid-run on a sunlit walkway with diagonal shadows; vivid red knee-high socks with fluttering ribbon, glossy black Mary Jane kitten heels; elegant and playful mood; strong motion blur, shallow depth of field, warm film grain, high contrast, minimalist background." ``` ``` "A high-energy action motion sequence: capture a subject mid-movement as if frozen in time, then released back into motion. Begin with a moment of suspension — the body in mid-air, hair and fabric pulled upward — then flow into the follow-through of impact or landing. Cinematic realism with strong directional light, high shutter speed look for clarity, natural motion blur trailing behind the limbs. Camera: dynamic tracking at 35mm, shifting from a low angle to eye level, emphasizing intensity. The sequence should feel like a compressed burst of energy — one decisive action visualized across just a few seconds." ``` ``` "A white deer stands on a rocky hilltop under a dark blue sky. The camera zooms rapidly towards the deer's face. The deer stares directly at the camera, its white fur distinct against the rocks. Shadows play across its antlers as the camera moves closer." ``` *** ### **Parameters** | Name | Type | Default | Notes | | -------------- | ------ | --------------- | -------- | | `Prompt` | Text | — | Required | | `Duration` | Select | 5s | 5s, 10s | | `Aspect Ratio` | - | Portrait (9:16) | - | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------- | --------------- | --------------------- | | Text → Video | \~3 minutes | Prompt | Short cinematic clips | | Image → Video | \~3 minutes | Image, Prompt | Animate assets | *** ### **Output options** | Option | Values / notes | | ------------ | -------------- | | Duration | 5s, 10s | | Aspect ratio | 9:16 | *** ### **Prompt tips** * Describe camera motion explicitly; Kling responds well to motion tokens. * Use short duration for cost/time efficiency. *** # Kling O1 (by Kuaishou) ### Overview A novel type of video generation model with unique capabilities. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Aspect ratios:** Multiple options available *** ### What it's great for * Creative experimentation. * Unique visual styles. * Novel video generation approaches. * Artistic and experimental content. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | ---------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | Multiple options | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | -------------------- | | Text → Video | \~150s | Prompt | Creative generation | | Image → Video | \~150s | Image, Prompt | Animated experiments | *** ### Prompt tips * Novel approach to video generation. * Good for experimental and creative projects. * Explore unique visual styles. *** # Seedance (by Bytedance) ### **Overview** Multi-shot text or image → video model for coherent, cinematic multi-shot output. *** ### **Quick facts** * **Modes:** Text → Video · Image → Video. * **Default clip length:** 5 seconds (3s–12s selectable) * **Aspect ratios:** 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 * **Training / licensing:** Proprietary model by Bytedance, licensed for commercial use within FLORA *** ### **What it’s great for** * Multi-shot narrative sequences and longer-form short clips. * High-coherence motion and smooth playback. * Storytelling use cases requiring multiple shots. *** #### **Copy-and-paste prompts** ``` Three-shot sequence: establish city skyline, mid shot of protagonist walking, close-up face — cinematic pacing — 10s total ``` ``` Image→Video: animate product spin across three cuts, consistent lighting and reflections ``` ``` Single-shot 1080p: slow crane up from street to rooftop, ambient city ambience ``` *** ### **Parameters** | Name | Type | Default | Notes | | -------------- | ------ | ------- | -------------------------------------------- | | `Prompt` | Text | — | Required | | `Duration` | Select | 10s | 3s–12s options (higher values increase cost) | | `Resolution` | Select | 1080p | 480p, 720p, 1080p | | `Seed` | Seed | Random | Optional | | `Fixed Camera` | Select | No | Yes, No | *** ### **Modes** | Mode | Estimated time | Required inputs | Typical use | | ------------------------- | -------------- | --------------- | ------------------------------ | | Text → Video (multi-shot) | \~2 minutes | Prompt | Narrative multi-shot sequences | | Image → Video | \~2 minutes | Image, Prompt | Animate assets across shots | *** ### **Output options** | Option | Values / notes | | ---------- | ----------------------------------------- | | Resolution | 1280×720, 1920×1080 | | Duration | 3s, 4s, 5s, 6s, 7s, 8s, 9s, 10s, 11s, 12s | | Formats | MP4 (H.264 + AAC) | | Notes | Designed for narrative and smooth motion | *** ### **Prompt tips** * Break multi-shot intent into clear shot descriptions (shot 1:, shot 2:, shot 3:). * Use pacing terms like “slow”, “cut-to”, “dolly”, and exact durations. *** # LTX-2 Pro (by Lightricks) ### Overview Fast video model with built-in audio generation and up to 4K resolution. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Resolutions:** 1080p, 1440p, 2160p (4K) * **Audio:** Built-in audio generation * **Aspect ratios:** 16:9 *** ### What it's great for * Quick video generation with audio. * High-resolution outputs up to 4K. * Fast turnaround for professional content. * Videos with synchronized sound. *** ### Parameters | Control | Type | Default | Notes | | -------------- | -------- | ------- | ----------------------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | 16:9 | | Duration | Dropdown | 6s | 6s, 8s (1.33x), 10s (1.66x) | | Resolution | Dropdown | 1080p | 1080p, 1440p (2x), 2160p (4x) | | FPS | Dropdown | 25 | 25, 50 | | Generate Audio | Toggle | On | Built-in audio generation | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | --------------------- | | Text → Video | \~60s | Prompt | Fast video with audio | | Image → Video | \~60s | Image, Prompt | Quick animation | *** ### Prompt tips * Model supports up to 4K resolution. * Audio generation is automatic. * Great balance of speed and quality. *** # Luma Ray 2 (by Luma) ### Overview Large-scale model for realistic, natural, and coherent motion. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Resolutions:** Multiple options available * **Aspect ratios:** Multiple options available *** ### What it's great for * Lifestyle, nature, and travel footage. * Drone-style glides and smooth camera flows. * Fashion runways and B-roll sequences. * Realistic, natural motion. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | -------------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | Multiple options | | Resolution | Dropdown | — | Multiple options | | Duration | Dropdown | — | Multiple options | | Loop | Toggle | Off | Create looping video | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | -------------------- | | Text → Video | \~200s | Prompt | Natural motion video | | Image → Video | \~200s | Image, Prompt | Animated sequences | *** ### Prompt tips * Excels at natural, realistic motion. * Great for cinematic camera movements. * Ideal for lifestyle and nature content. *** # Pika (by Pika) ### Overview General model with good prompt adherence and expressive faces. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Resolutions:** Multiple options available * **Aspect ratios:** Multiple options available *** ### What it's great for * Avatar-led explainers. * Talking-head shorts. * Expressive facial performances. * Character-focused content. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | ---------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | Multiple options | | Duration | Dropdown | — | Multiple options | | Resolution | Dropdown | — | Multiple options | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------- | | Text → Video | \~90s | Prompt | Expressive videos | | Image → Video | \~90s | Image, Prompt | Character animation | *** ### Prompt tips * Great for character-focused content. * Strong facial expression capabilities. * Good prompt adherence. *** # Minimax Hailuo (by Minimax) ### Overview Powerful, motion-heavy model with nuanced prompt comprehension. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Aspect ratios:** Multiple options available *** ### What it's great for * Narrative shorts (30-60 seconds). * Coherent character arcs. * Evolving lighting cues. * Motion-heavy content. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | ---------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | Multiple options | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------ | | Text → Video | \~150s | Prompt | Narrative content | | Image → Video | \~150s | Image, Prompt | Story-driven video | *** ### Prompt tips * Excels at narrative content. * Good for character-driven stories. * Handles complex lighting well. *** # Tencent Hunyuan (by Tencent) ### Overview High-quality and customizable video model with video-to-audio pairing. *** ### Quick facts * **Modes:** Text → Video · Image → Video * **Resolutions:** Multiple options available * **Aspect ratios:** Multiple options available *** ### What it's great for * Multi-locale ad campaigns. * Culturally nuanced content. * Language-specific versions. * Synced soundtrack stubs. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | ---------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | Multiple options | | Resolution | Dropdown | — | Multiple options | | Pro Mode | Toggle | Off | Enhanced quality | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | ------------------ | | Text → Video | \~180s | Prompt | Customizable video | | Image → Video | \~180s | Image, Prompt | Localized content | *** ### Prompt tips * Good for international campaigns. * Supports audio pairing. * Highly customizable outputs. *** # WAN 2.6 (by Alibaba) ### Overview Latest WAN model with multi-shot support and high-resolution outputs. *** ### Quick facts * **Modes:** Text → Video · Image → Video · Video → Video * **Resolutions:** Multiple options available * **Features:** Multi-shot support *** ### What it's great for * Complex multi-shot sequences. * High-resolution video outputs. * Versatile video generation. * Professional content creation. *** ### Parameters | Control | Type | Default | Notes | | ------------ | -------- | ------- | ----------------- | | Prompt | Text | — | Required | | Aspect Ratio | Dropdown | 16:9 | Multiple options | | Resolution | Dropdown | — | Multiple options | | Duration | Dropdown | — | Multiple options | | Multi-shots | Toggle | Off | Enable multi-shot | *** ### Modes | Mode | Estimated time | Required inputs | Typical use | | ------------- | -------------: | --------------- | -------------------- | | Text → Video | \~150s | Prompt | Multi-shot sequences | | Image → Video | \~150s | Image, Prompt | Animated content | | Video → Video | \~150s | Video, Prompt | Video transformation | *** ### Prompt tips * Supports multi-shot video generation. * Good for complex sequences. * High-resolution capable. *** # Marey (by Moonvalley) ### **Overview** High-quality video generation and motion-transfer model for short cinematic clips. *** ### **Quick facts** * **Modes:** Text → Video · Image → Video · Motion / Pose Transfer (video → video) * **Default clip length:** 5s (supports 10s) * **Aspect ratios:** Auto, 16:9, 4:3, 1:1, 3:4, 9:16 * **Licensing:** Trained only on fully licensed, high-resolution footage (no scraped or user-submitted data) *** ### **What it’s great for** * Short cinematic pieces and motion studies * Turning still images into motion (image → video) * Transferring motion or pose from reference videos to targets *** ### **Example outputs** {% hint style="info" %} Hero frame — static hero still {% endhint %}
{% hint style="info" %} 5s cinematic clip (image → video) {% endhint %} {% embed url="" %} {% hint style="info" %} 10s pose-transfer sequence {% endhint %} {% embed url="" %} #### **Copy-and-paste prompts** ``` A lone astronaut walking along a moonlit cliff, slow dolly in, cinematic lighting, film grain — 5s — 16:9 ``` ``` Marble statue by the sea slowly coming to life, subtle camera orbit, warm cinematic grade — 10s — 1:1 ``` ``` Close-up portrait, gentle breathing and head turn, handheld camera feel, soft rim light — 5s — 4:3 ``` ``` Motion transfer: apply running motion from reference video to a still image of a red bicycle in an empty city square — 5s — 16:9 ``` *** ### **Parameters** | Name | Type | Default | Notes | | -------------- | ------ | ---------------- | --------------------------------------------------------------- | | `Prompt` | Text | — | Natural-language scene, motion, style description (required) | | `Duration` | Select | 5s | 5s (default), 10s (cost multiplier 2×) | | `Aspect Ratio` | Select | Landscape (16:9) | Landscape (16:9), Landscape (4:3), Square (1:1), Portrait (3:4) | | `Seed` | Seed | Random | Optional for deterministic / repeatable outputs | *** ### **Modes & endpoints** | Mode | Estimated time | Required inputs | Typical use | | ------------------------------- | -------------- | ----------------------------------- | ------------------------------------------ | | Text → Video (t2v) | \~300s | Prompt | Short cinematic clip from text | | Image → Video (i2v) | \~420s | Image, Prompt | Animate still images | | Video → Video (motion transfer) | \~400s | Video, Prompt, Optional First Frame | Transfer motion from reference to target | | Video → Video (pose transfer) | \~300–400s | Video, Prompt | Transfer pose/gesture from reference video | *** ### **Output options** | Option | Values / notes | | ------------ | ----------------------------------------------------------------------------------------------- | | Duration | 5s (default), 10s (longer clips may cost more) | | Aspect ratio | 16:9 (1920×1080), 4:3 (1536×1152), 1:1 (1152×1152), 3:4 (1152×1536), 9:16 (1080×1920) | | Notes | Marey generates up to \~5 seconds of consistent 24 FPS footage per clip in its public offering. | *** ### **Prompt tips** * Specify **motion** explicitly: e.g., “slow dolly in”, “camera orbit”, “subtle head turn”. * Mention **shot specifics** when important: camera framing, lens feel, lighting, and duration. * Use seed for repeatable variants or test different seeds to explore diversity. *** ### **Safety** Marey is built for professional filmmaking workflows and emphasizes ethical sourcing and legal clarity for commercial use. *** # Audio Models Here are all the audio models currently available in FLORA. ## Speech & Voice These models generate or transcribe speech.
ModelEst. time (s)DescriptionBest for
ElevenLabs Multilingual v215Natural text-to-speech with multilingual voice synthesis.Voiceovers, narration, multilingual content, accessible media.
ElevenLabs Scribe v210Transcribe speech to text.Transcription, subtitles, meeting notes, content indexing.
Gemini 3.1 Flash TTS10Text to speech with prompted expressiveness.Expressive voiceovers, character dialog, and stylized narration.
## Sound Effects & Music These models generate sound effects and music from text prompts.
ModelEst. time (s)DescriptionBest for
ElevenLabs Music v130Generate music using a text prompt.Background tracks, musical stings, and prompt-driven composition.
ElevenLabs Sound Effects15Generate sound effects and Foley from text descriptions.Foley, ambient soundscapes, UI sounds, game audio.
# How Pricing Works {% hint style="success" %} **Launch bonus extended through August 31, 2026.** Every paid plan gets extra usage during launch. See the [Launch Bonus](/plans-and-billing/launch-bonus) page for full details. {% endhint %} **Starter and above** include a monthly usage budget (denominated in dollars) for **text, image, video, audio, and other models** in the app. One usage bar. No per-generation math in the flow. Usage resets each month — it doesn't carry over. The **Free** plan is for trying the canvas: **text and image models** only (no video models in the app), **limited complimentary generations** for exploration — not a monthly dollar allowance — plus **no FLORA API or MCP** (those start on **Starter**). See [flora.ai/pricing](https://flora.ai/pricing) for the current free-generation cap and footnotes. ## Plans | Plan | Price | Monthly usage per seat | Launch bonus | Total during launch | Max seats | | ---------- | ------- | ---------------------- | ------------ | ------------------- | --------- | | Free | $0 | — | — | — | 1 | | Starter | $18/mo | $18 | +$12 | $30 | 8 | | Pro | $50/mo | $50 | +$50 | $100 | 8 | | Max | $200/mo | $200 | +$100 | $300 | 8 | | Enterprise | Custom | Custom | — | — | Custom | {% hint style="info" %} **Plan features compared.** The table above covers price, usage, and seats — for the full feature comparison (workspace controls, team features, support tiers, etc.), see [flora.ai/pricing](https://flora.ai/pricing). {% endhint %} Annual plans save 20% on **Starter, Pro, and Max**. Those plans share the same published per-generation rates for every model your plan can access in the app, and **Starter and above** include **API & MCP**. **Free** stays on **text and image** in the product, **explore-only generation limits**, and **no API or MCP**. FAUNA is free on every plan, always — it doesn't count against your budget. ## How Usage Works **On paid plans,** your monthly budget is denominated in dollars. When you run a generation, the model's cost is deducted from your budget. You see a simple percentage bar in the header — by default, no dollar amounts in the generation flow, no per-generation math. If you'd rather see exact costs, toggle **Preferences → Show generation costs in dollars** to replace the flower meter with the per-run dollar amount on every model and Run button. See [Model Pricing](/plans-and-billing/model-pricing#show-exact-dollar-amounts). **What counts as usage.** Any generation from an AI model — images, video, audio, text — based on the model's published cost. FAUNA messages and interactions don't count against your budget on any plan. **Where to see your usage.** A live percentage bar lives in the app header. Open **Settings → People & Seats** and scroll to **Usage this period** to see included usage for the current billing cycle, how much is left, when your budget **resets**, and any overage for the period.
Usage this period on People and Seats showing included usage, reset date, and overage

People & Seats → Usage this period

**Usage History** (under **Plan & Usage**) shows exact dollar cost per generation, per project, and per member — useful for client billing or team budgets. **Model costs.** A public model pricing page lists the rate for every model we support, so you can compare if you want to plan around a specific workflow. ## Teams Every paid plan includes team collaboration, and each seat adds both cost and budget based on your plan's per-seat rates. **Example (Pro plan):** * Pro costs **$50 per seat/month** * Pro includes **$50 usage per seat/month** * So a 4-person Pro workspace = **4 x $50 = $200/month** and **4 x $50 = $200/month shared budget** During the [launch bonus](/plans-and-billing/launch-bonus) (through August 31, 2026), that 4-person Pro workspace's shared budget doubles to **$400/month**. Admins have control over how usage flows inside a workspace: * **Pooled or per-user.** By default, everyone draws from the same shared pool. Admins can switch to per-user budgets at any time. * **Per-member caps.** Set a dollar cap on any individual member to prevent one person from draining the pool. When a personal limit is set, that member's usage indicator in the app header shows their individual usage percentage (or "Limit reached" when the cap is hit) with a tooltip explaining the admin-set limit. Self-serve plans go up to 8 seats. Beyond that, [Enterprise](#enterprise) gives you a custom seat count plus admin, security, and advanced features. ## Extra Usage (On-Demand Spending) When your monthly budget runs out, you have two options: upgrade your plan, or turn on on-demand spending (called **Overage** in workspace settings). On-demand is: * **Opt-in.** Disabled by default. Workspace admins enable it in **Settings → People & Seats**. * **Capped.** You set a **max overage** limit for the billing period so there are no runaway charges. * **Card-required (self-serve).** A payment method must be on file. Enterprise accounts on invoice billing follow your contract instead. * **Same price as included usage.** No markup — you pay the published model rates. ### Your max overage cap The max overage setting is the most extra usage you allow **in the current billing period** — not just what is unpaid at the moment. * Flora tracks **total overage this period** (everything above your included budget and wallet balance, whether or not it has been invoiced yet). * When total overage would exceed your cap, **new generations stop** until the next billing period. * When total overage **reaches** your cap, any remaining balance is **billed** and overage stops for the rest of the period. If your cap is **lower** than your workspace's monthly subscription total, the cap controls both spending and billing. If your cap is **higher**, you can still be charged earlier when other thresholds are met (see below). Paying an overage charge does **not** reset your cap for the period — only your subscription renewal starts a fresh overage tally. ### When self-serve workspaces are charged For Starter, Pro, and Max workspaces paying by card, overage is **not** held until the end of the billing cycle. Charges happen when enough unpaid overage has built up: 1. **Max overage reached** — total overage this period hits the limit you set. 2. **Large unpaid balance** — unpaid overage reaches **$200** (a system safety threshold). 3. **Subscription-sized chunk** — unpaid overage reaches your workspace's **monthly subscription total** (plan price × editor seats; for annual billing, Flora uses the monthly equivalent). Example: one editor on Starter ($18/mo) is charged when about **$18** of overage is unpaid; four editors on Pro ($50 × 4 = $200/mo) is charged when about **$200** of overage is unpaid. 4. **Daily balance check** — if unpaid overage is still **more than $5** but has not hit the thresholds above, Flora attempts to collect it once per day (around 8:15 AM UTC). You may see **multiple charges in one period** as you keep generating — especially when your max overage is higher than your subscription total. Each charge covers what was unpaid at that moment; your period total keeps counting toward your cap. Charges appear on your card through Stripe. Open **People & Seats → Manage & Invoices** to view invoices and payment history. If a subscription payment fails, **new overage charges pause** until billing is healthy again. Existing prepaid or wallet balance is not removed. ### Enterprise overage billing Enterprise overage follows your contract. Most enterprise workspaces receive **invoices** (not automatic card charges) on a **monthly or quarterly** schedule tied to the contract start date, with payment terms set in your agreement (commonly net 60). The **People & Seats** overage settings show the next scheduled invoice date when invoice billing applies. Some enterprise setups use the same threshold rules as self-serve, but still bill via invoice rather than an immediate card charge. Your account team can confirm which model applies to your workspace. ## Enterprise For organizations that need a custom seat count, SSO, admin controls, audit logs, invoice billing, and the rest. Negotiated annually. [Contact us](https://flora.ai/contact) to start a conversation. *** ## Further Reading * [Launch Bonus](/plans-and-billing/launch-bonus) — extra usage through August 31, 2026 * [Pricing FAQ](/plans-and-billing/faq) — common questions about plans, usage, and billing * [Manage Subscription](/plans-and-billing/manage-subscription) — change, upgrade, or cancel your plan * [flora.ai/pricing](https://flora.ai/pricing) — full feature comparison across plans ## Questions? * General: * Enterprise: [flora.ai/contact](https://flora.ai/contact) # Model Pricing Per-generation cost for every model FLORA supports. On **Starter and above**, each generation deducts its dollar cost from your workspace's monthly usage budget — see [How Pricing Works](/plans-and-billing/pricing) for plan budgets, or the [Pricing FAQ](/plans-and-billing/faq) for billing details. The **Free** plan uses **limited complimentary generations** instead of a monthly dollar pool (see [flora.ai/pricing](https://flora.ai/pricing)). * **Failed generations don't count.** * **FAUNA is free** on every plan and doesn't draw from your budget. * **On-demand spending uses the same rates** — no markup for going over your monthly budget. {% hint style="success" %} **Launch bonus extended through August 31, 2026.** Every paid plan has extra usage on top of its baseline. See the [Launch Bonus](/plans-and-billing/launch-bonus) page. {% endhint %} ## Reading the usage meter Every model has a usage meter — a small row of flowers indicating roughly what a generation will cost. More flowers, higher cost. When the cost band is **$2 or more**, the meter shows **three flowers plus a “+”**.
MeterCost per generation
One flowerLess than $0.10
Two flowers$0.10 – $1
Three flowers$1 – $2
Three flowers with a plus sign$2 or more
Where you'll see it:
Usage meter shown next to a model
### Show exact dollar amounts If you'd rather see the precise cost than read the flower meter, turn on **Preferences → Show generation costs in dollars**. Open Preferences from the workspace menu — the toggle is global and applies to every project in your account.
Preferences panel with Show generation costs in dollars toggled on

Preferences → Show generation costs in dollars

With it enabled: * Every model in the picker shows its per-generation cost (e.g. `$0.052`) in place of the flower meter.
Model list showing FREE and dollar costs next to each model

Per-model cost shown in the picker

* The **Run** button on each node shows the exact dollar amount for the run, accounting for batch count and any per-run modifiers.
Run button showing the exact dollar cost for the current run

Run button with the exact cost for the current run

For the exact rate per model, see the tables below. {% hint style="info" %} **Pricing is dynamic — your cost changes with the parameters you choose**, like resolution and clip length. For Image and Video models, the **Cost per generation** column is the price at each model's **Default settings**, and **Price range** shows how low and high the cost can go as you change those settings (each end notes the settings that produce it). For the price of every individual combination, see [Pricing Breakdown by Parameter](/plans-and-billing/parameter-pricing). **The exact cost for your specific parameters is always shown in the app before you generate.** {% endhint %}
Generate panel showing duration, resolution, and other settings with the Cost row displaying the exact dollar amount for the current configuration

Cost updates in the Generate panel as you change model settings; tables below list default-settings prices and ranges

## Text Models | Model | Provider | Cost per generation | | ----------------- | --------- | ------------------- | | GPT 4o Mini | OpenAI | $0.0018 | | Claude Sonnet 4.5 | Anthropic | $0.0072 | | Claude Sonnet 4.6 | Anthropic | $0.0072 | | Gemini 3 Flash | Google | $0.0072 | | Claude Opus 4.5 | Anthropic | $0.0090 | | Claude Opus 4.6 | Anthropic | $0.0090 | | Claude Opus 4.7 | Anthropic | $0.0090 | | Claude Sonnet 5 | Anthropic | $0.0108 | | Claude Opus 4.8 | Anthropic | $0.0135 | | Claude Opus 5 | Anthropic | $0.0135 | | Gemini 2.5 Pro | Google | $0.0216 | | GPT-5 | OpenAI | $0.0234 | | GPT-5.1 | OpenAI | $0.0234 | | Gemini 3.1 Pro | Google | $0.0252 | | GPT-5.2 | OpenAI | $0.0324 | | GPT-5.4 | OpenAI | $0.0450 | | GPT-5.5 | OpenAI | $0.1350 | | Claude Fable 5 | Anthropic | $0.7200 | | o3 Deep Research | OpenAI | $0.8102 | ## Image Models Includes generation, editing, inpainting, character, vector, upscaling, background removal, and training utilities. | Model | Provider | Cost per generation | Default settings | Price range | | ------------------------------ | ----------------- | ------------------- | ------------------------- | ------------------------------------- | | Remove Background | BiRefNet | $0.0018 | — | — | | Z-Image Turbo | Alibaba | $0.0054 | — | — | | Flux 2 | Black Forest Labs | $0.0094 | — | — | | Flux 2 Turbo | Black Forest Labs | $0.0096 | — | — | | Flux 2 Klein 4B | Black Forest Labs | $0.0120 | — | — | | Flux 2 Klein 9B | Black Forest Labs | $0.0132 | — | — | | Riverflow 2.0 Fast | Sourceful | $0.0216 | — | — | | Luma Photon | Luma | $0.0234 | — | — | | Seedream 3.0 | ByteDance | $0.0252 | — | — | | Seedream 4.0 | ByteDance | $0.0252 | — | — | | Wan 2.2 | Alibaba | $0.0252 | — | — | | Kling O1 | Kling | $0.0270 | — | — | | Flux Dev | Black Forest Labs | $0.0306 | — | — | | Seedream 4.5 | ByteDance | $0.0336 | — | — | | Flux 2 Pro | Black Forest Labs | $0.0360 | — | — | | Krea 2 Medium | Krea | $0.0360 | — | — | | Qwen Image Edit | Alibaba | $0.0360 | auto | from $0.02 (21:9) | | Qwen Image Edit Plus | Alibaba | $0.0360 | auto | from $0.02 (21:9) | | Flux Redux | Black Forest Labs | $0.0370 | — | — | | Flux 2 Flex | Black Forest Labs | $0.0390 | — | — | | Flux Canny | Black Forest Labs | $0.0396 | — | — | | Krea 2 References Medium | Krea | $0.0420 | — | — | | Qwen Image 2.0 | Alibaba | $0.0420 | auto | from $0.02 (21:9) | | Qwen Image Edit 2511 Angles | Alibaba | $0.0420 | — | — | | Flux Depth | Black Forest Labs | $0.0424 | — | — | | Grok Imagine | xAI | $0.0424 | — | — | | Seedream 5.0 Lite | ByteDance | $0.0424 | — | — | | Stable Diffusion 3.5 | Stability AI | $0.0424 | — | — | | Nano Banana 2 Lite | Google | $0.0428 | — | — | | GPT Image | OpenAI | $0.0450 | — | — | | Nano Banana | Google | $0.0468 | — | — | | Imagen 4 | Google | $0.0480 | — | — | | Recraft V4.1 | Recraft | $0.0480 | — | — | | Recraft V4.1 Utility | Recraft | $0.0480 | — | — | | Uni-1 | Luma | $0.0480 | — | — | | Imagen 3 | Google | $0.0486 | — | — | | Imagen 3 Inpainting | Google | $0.0486 | — | — | | Imagen 3 Outpainting | Google | $0.0486 | — | — | | Recraft V3 | Recraft | $0.0486 | — | — | | Recraft V4 | Recraft | $0.0486 | — | — | | Flux 2 Max | Black Forest Labs | $0.0546 | — | — | | Grok Imagine Quality | xAI | $0.0600 | 1k | up to $0.30 (2k) | | Flux Pro 1.1 | Black Forest Labs | $0.0720 | — | — | | Ideogram 3.0 | Ideogram | $0.0720 | — | — | | Ideogram 4.0 | Ideogram | $0.0720 | rendering speed DEFAULT | $0.04 (TURBO) → $0.12 (QUALITY) | | Krea 2 Large | Krea | $0.0720 | — | — | | Nano Banana 2 | Google | $0.0720 | 1K · web search off | up to $0.16 (4K · web search on) | | Krea 2 References Large | Krea | $0.0780 | — | — | | Flux Kontext Max | Black Forest Labs | $0.0810 | — | — | | Ideogram Character | Ideogram | $0.0900 | — | — | | Recraft V4.1 Vector | Recraft | $0.0960 | — | — | | Recraft V4 Vector | Recraft | $0.0964 | — | — | | Uni-1 Max | Luma | $0.1200 | — | — | | Enhancor V1 | Enhancor | $0.1206 | enhancement standard | up to $0.33 (heavy) | | Magnific Creative Upscaler | Magnific | $0.1206 | scale factor 2x | up to $0.96 (16x) | | Magnific Precision Upscaler | Magnific | $0.1206 | — | — | | Magnific Precision Upscaler V2 | Magnific | $0.1206 | — | — | | Riverflow 2.0 Pro | Sourceful | $0.1620 | 2K | up to $0.32 (4K) | | Riverflow 2.0 Pro Inpainting | Sourceful | $0.1620 | 2K | up to $0.32 (4K) | | GPT Image 1.5 | OpenAI | $0.1680 | — | — | | Seedream 5 Pro | ByteDance | $0.1699 | 2k | from $0.08 (1k) | | Nano Banana Pro | Google | $0.1800 | 1K | up to $0.36 (4K) | | Nano Banana Pro Inpainting | Google | $0.1800 | 1K | up to $0.36 (4K) | | Topaz Upscaler | Topaz | $0.1908 | — | — | | GPT Image 2 | OpenAI | $0.2635 | 1k · quality high | $0.02 (1k · low) → $0.87 (4k · high) | | GPT Image 1.5 Inpainting | OpenAI | $0.2772 | — | — | | Riverflow 2.5 Pro | Sourceful | $0.2832 | 2K · thinkingLevel medium | $0.27 (2K · low) → $0.87 (4K · xhigh) | | Arrow 1.0 | QuiverAI | $0.2880 | — | — | | Arrow 1.0 References | QuiverAI | $0.2880 | — | — | | Arrow 1.1 | QuiverAI | $0.2880 | — | — | | Arrow 1.1 References | QuiverAI | $0.2881 | — | — | | Recraft V4 Pro | Recraft | $0.2881 | — | — | | Recraft V4.1 Pro | Recraft | $0.3000 | — | — | | Recraft V4.1 Utility Pro | Recraft | $0.3000 | — | — | | Topaz Generative Upscaler | Topaz | $0.3120 | — | — | | Reve 2.1 | Reve | $0.3150 | — | — | | Arrow 1.1 Max | QuiverAI | $0.3151 | — | — | | Arrow 1.1 Max References | QuiverAI | $0.3151 | — | — | | Enhancor V4 | Enhancor | $0.3367 | fast mode on | up to $0.96 (fast mode off) | | Recraft V4 Pro Vector | Recraft | $0.3600 | — | — | | Recraft V4.1 Pro Vector | Recraft | $0.3600 | — | — | | Enhancor V3 | Enhancor | $0.4663 | 1024 | up to $0.84 (3072) | | Lora Trainer | Stability AI | $1.8005 | — | — | ## Video Models Includes generation, editing, motion control, lipsync, animation, video upscaling, and background removal. | Model | Provider | Cost per generation | Default settings | Price range | | --------------------------------------- | ------------ | ------------------- | ---------------- | ---------------------------------------------------- | | VEED Subtitles | VEED | $0.0240 | — | — | | Topaz Upscaler | Topaz | $0.0480 | — | — | | Animatediff | Stability AI | $0.0900 | — | — | | Gemini-Omni-Flash | Google | $0.1296 | — | — | | Veed Background Removal | VEED | $0.1350 | — | — | | Frame Morphing | Stability AI | $0.1720 | — | — | | Pika | Pika | $0.1920 | 1080p · 5s | $0.08 (720p · 5s) → $0.42 (1080p · 10s) | | Minimax Hailuo | MiniMax | $0.2280 | — | — | | Luma Ray 2 Flash | Luma | $0.2881 | 720p · 5s · 16:9 | $0.09 (540p · 5s · 1:1) → $0.68 (720p · 9s · 21:9) | | Kling 2.5 Turbo Pro | Kling | $0.2940 | 5s | up to $0.59 (10s) | | Seedance 1.5 Pro | ByteDance | $0.3124 | 720p · 5s | $0.12 (480p · 4s) → $0.75 (720p · 12s) | | Kling Pro 1.6 | Kling | $0.3304 | 5s | up to $0.66 (10s) | | LTX-2 Pro | Lightricks | $0.4321 | 1080p · 6s | up to $2.87 (2160p · 10s) | | Kling 2.1 Pro | Kling | $0.4500 | 5s | up to $0.90 (10s) | | Kling O1 | Kling | $0.4704 | 5s | up to $0.94 (10s) | | Kling O1 References | Kling | $0.4704 | 5s | up to $0.94 (10s) | | Kling O3 Pro References | Kling | $0.4704 | 1080p · 5s | $0.28 (1080p · 3s) → $2.70 (4k · 15s) | | Kling O3 Standard | Kling | $0.4704 | 5s | $0.28 (3s) → $1.41 (15s) | | Minimax Hailuo 02 Pro | MiniMax | $0.4800 | — | — | | WAN 2.2 Animate Move | Alibaba | $0.4800 | 720p | from $0.24 (480p) | | WAN 2.2 Animate Replace | Alibaba | $0.4800 | 720p | from $0.24 (480p) | | Tencent Hunyuan | Tencent | $0.4807 | pro mode off | up to $0.93 (pro mode on) | | Veo 3.1 Frames Lite | Google | $0.4807 | 720p · 8s | $0.24 (720p · 4s) → $1.12 (4k · 8s) | | Veo 3.1 Lite | Google | $0.4807 | 8s | from $0.24 (4s) | | WAN 2.2 | Alibaba | $0.4817 | 720p · 5s | $0.32 (480p · 5s) → $0.87 (720p · 8s) | | Kling 3.0 Standard | Kling | $0.5292 | 5s | $0.32 (3s) → $1.59 (15s) | | Kling O3 Standard Reference | Kling | $0.5292 | 5s | $0.32 (3s) → $1.59 (15s) | | Kling O3 Standard References | Kling | $0.5292 | 5s | $0.32 (3s) → $1.59 (15s) | | Minimax Hailuo 2.3 Pro | MiniMax | $0.5761 | — | — | | Kling O3 Pro | Kling | $0.5880 | 1080p · 5s | $0.35 (1080p · 3s) → $3.38 (4k · 15s) | | Lucy Edit Pro | Decart AI | $0.6000 | 720p | from $0.30 (480p) | | Mirelo SFX 1.5 | Mirelo | $0.6000 | — | — | | Kling Pro 1.5 | Kling | $0.6005 | — | — | | LTX-2 Retake | Lightricks | $0.6005 | — | — | | WAN 2.5 | Alibaba | $0.6005 | 720P · 5s | up to $1.80 (1080P · 10s) | | WAN 2.6 | Alibaba | $0.6005 | 720P · 5s | up to $2.70 (1080P · 15s) | | WAN 2.7 | Alibaba | $0.6005 | 720P · 5s | up to $2.70 (1080P · 15s) | | Veed Lipsync | VEED | $0.6240 | — | — | | Kling 3.0 Standard (Turbo) | Kling | $0.6720 | 5s | $0.40 (3s) → $2.02 (15s) | | Kling 2.6 Pro | Kling | $0.6762 | 5s | up to $1.35 (10s) | | Kling Avatar v2 Pro | Kling | $0.6900 | — | — | | Kling 3.0 Pro | Kling | $0.7056 | 1080p · 5s | $0.42 (1080p · 3s) → $3.53 (4k · 15s) | | Kling O1 Edit | Kling | $0.7056 | — | — | | Kling O1 Reference | Kling | $0.7056 | 5s | up to $1.41 (10s) | | Kling O3 Pro Reference | Kling | $0.7056 | 5s | $0.42 (3s) → $2.12 (15s) | | Lipsync 2 Pro | Sync | $0.7200 | — | — | | Runway Gen-4.5 | Runway | $0.7202 | 5s | up to $1.44 (10s) | | Seedance 1.0 Pro | ByteDance | $0.7445 | 1080p · 5s | $0.09 (480p · 3s) → $1.79 (1080p · 12s) | | Kling 3.0 Standard Motion Control | Kling | $0.7560 | — | — | | Aurora | Creatify | $0.8400 | 720p | from $0.42 (480p) | | Bria Video Upscaler | Bria | $0.8400 | — | — | | Happy Horse 1.0 | Alibaba | $0.8400 | 720p · 5s | $0.50 (720p · 3s) → $3.36 (1080p · 15s) | | Happy Horse 1.0 Edit | Alibaba | $0.8400 | 720p | up to $1.68 (1080p) | | Happy Horse 1.0 References | Alibaba | $0.8400 | 720p · 5s | $0.50 (720p · 3s) → $3.36 (1080p · 15s) | | Kling 3.0 Pro (Turbo) | Kling | $0.8400 | 5s | $0.50 (3s) → $2.52 (15s) | | Kling O3 Pro Edit | Kling | $0.8400 | — | — | | Kling O3 Standard Edit | Kling | $0.8400 | — | — | | Seedance 2.0 Fast | ByteDance | $0.8514 | 720p · 5s | $0.22 (480p · 4s) → $2.55 (720p · 15s) | | Seedance 2.0 Reference (Fast) | ByteDance | $0.8514 | 720p · 5s | $0.22 (480p · 4s) → $2.55 (720p · 15s) | | Seedance 2.0 Reference (Fast) (Runware) | ByteDance | $0.8514 | — | — | | Luma Ray 2 | Luma | $0.8525 | 720p · 5s · 16:9 | $0.27 (540p · 5s · 1:1) → $2.01 (720p · 9s · 21:9) | | Magnific Creative Turbo Video Upscaler | Magnific | $0.8822 | — | — | | Fabric 1.0 | VEED | $0.9000 | 720p | from $0.50 (480p) | | Seedance 2.0 Mini | ByteDance | $0.9288 | 720p · 5s | $0.22 (480p · 4s) → $3.05 (720p · 15s) | | Seedance 2.0 | ByteDance | $1.0584 | 720p · 5s | $0.28 (480p · 4s) → $7.41 (4k · 15s) | | Seedance 2.0 Reference | ByteDance | $1.0584 | 720p · 5s | $0.28 (480p · 4s) → $7.41 (4k · 15s) | | Seedance 2.0 Reference (Runware) | ByteDance | $1.0584 | — | — | | Seedance 2.0 Reference Multi | ByteDance | $1.0584 | — | — | | Veo 2 | Google | $1.0802 | 5s | up to $1.86 (8s) | | Sync 3 | Sync | $1.1520 | — | — | | Magnific Creative Video Upscaler | Magnific | $1.2064 | — | — | | Magnific Precision Video Upscaler | Magnific | $1.2064 | — | — | | Grok Imagine | xAI | $1.3036 | 6s | $0.21 (1s) → $3.26 (15s) | | Grok Imagine References | xAI | $1.3036 | 6s | $0.21 (1s) → $3.26 (15s) | | Grok Imagine Edit | xAI | $1.3108 | — | — | | Kling 2.0 Master | Kling | $1.3810 | — | — | | Kling 2.1 Master | Kling | $1.3810 | 5s | up to $2.76 (10s) | | Sora 2 Pro | OpenAI | $1.4400 | 720p · 4s | up to $7.17 (1080p · 12s) | | Veo 3.1 (Fast) | Google | $1.4400 | 720p · 8s | $0.96 (720p · 4s) → $3.36 (4k · 8s) | | Veo 3.1 Frames (Fast) | Google | $1.4400 | 720p · 8s | $0.96 (720p · 4s) → $3.36 (4k · 8s) | | Kling 2.6 Pro Motion Control | Kling | $1.5994 | — | — | | Marey | Moonvalley | $1.8005 | 5s | up to $3.60 (10s) | | Kling 3.0 Pro Motion Control | Kling | $2.0160 | — | — | | Luma Modify Video | Luma | $2.1011 | — | — | | Marey Motion Transfer | Moonvalley | $2.4018 | — | — | | Marey Pose Transfer | Moonvalley | $2.4018 | — | — | | Seedance 2.5 | ByteDance | $2.4846 | 720p · 5s · 16:9 | $0.15 (480p · 4s · 1:1) → $15.68 (720p · 30s · 21:9) | | Veo 3 | Google | $3.8400 | — | — | | Veo 3.1 | Google | $3.8400 | 720p · 8s | $3.36 (720p · 4s) → $5.76 (4k · 8s) | | Veo 3.1 Frames | Google | $3.8400 | 720p · 8s | $3.36 (720p · 4s) → $5.76 (4k · 8s) | | Veo 3.1 Ingredients | Google | $3.8400 | 720p | up to $5.76 (4k) | ## Audio Models | Model | Provider | Cost per generation | | -------------------------- | ---------- | ------------------- | | ElevenLabs Scribe v2 | ElevenLabs | $0.0036 | | MMaudio v2 | BiRefNet | $0.0060 | | Gemini 3.1 Flash TTS | Google | $0.0126 | | ElevenLabs Multilingual v2 | ElevenLabs | $0.1206 | | ElevenLabs Sound Effects | ElevenLabs | $0.1440 | | ElevenLabs Music v1 | ElevenLabs | $0.3601 | *** ## Questions? * General: * Enterprise: [flora.ai/contact](https://flora.ai/contact) # Pricing Breakdown by Parameter Full price breakdown for every Image and Video model whose cost changes with resolution, duration, or other settings. For the at-a-glance list of every model's default-settings price, see [Model Pricing](/plans-and-billing/model-pricing). {% hint style="info" %} **Pricing is dynamic — your cost changes with the parameters you choose.** Each table below shows the price for every combination of cost-affecting settings, computed from the model's base rate and its published per-parameter modifiers. The exact cost for your specific parameters — including batch count — is always shown in the app before you generate. Models not listed here generate at a single flat rate (see [Model Pricing](/plans-and-billing/model-pricing)). {% endhint %} ## Video Models #### Pika Default: **$0.1920** — 1080p · 5s | Duration \ Resolution | 720p | 1080p *(default)* | | --------------------- | ----- | ----------------- | | 5s *(default)* | $0.08 | $0.19 | | 10s | $0.19 | $0.42 | #### Luma Ray 2 Flash Default: **$0.2881** — 720p · 5s · 16:9 | Duration \ Resolution | 540p | 720p *(default)* | | --------------------- | ----- | ---------------- | | 5s *(default)* | $0.16 | $0.29 | | 9s | $0.29 | $0.52 | * **aspect ratio:** 21:9 $0.3782 · 16:9 *(default)* $0.2881 · 4:3 $0.2161 · 1:1 $0.1621 · 3:4 $0.2161 · 9:16 $0.2881 · 9:21 $0.3782 #### Kling 2.5 Turbo Pro Default: **$0.2940** — 5s * **duration:** 5s *(default)* $0.2940 · 10s $0.5880 #### Seedance 1.5 Pro Default: **$0.3124** — 720p · 5s | Duration \ Resolution | 480p | 720p *(default)* | | --------------------- | ----- | ---------------- | | 4s | $0.12 | $0.25 | | 5s *(default)* | $0.16 | $0.31 | | 6s | $0.19 | $0.37 | | 7s | $0.22 | $0.44 | | 8s | $0.25 | $0.50 | | 9s | $0.28 | $0.56 | | 10s | $0.31 | $0.62 | | 11s | $0.34 | $0.69 | | 12s | $0.37 | $0.75 | #### Kling Pro 1.6 Default: **$0.3304** — 5s * **duration:** 5s *(default)* $0.3304 · 10s $0.6607 #### LTX-2 Pro Default: **$0.4321** — 1080p · 6s | Duration \ Resolution | 1080p *(default)* | 1440p | 2160p | | --------------------- | ----------------- | ----- | ----- | | 6s *(default)* | $0.43 | $0.86 | $1.73 | | 8s | $0.57 | $1.15 | $2.30 | | 10s | $0.72 | $1.43 | $2.87 | #### Kling 2.1 Pro Default: **$0.4500** — 5s * **duration:** 5s *(default)* $0.4500 · 10s $0.9000 #### Kling O1 Default: **$0.4704** — 5s * **duration:** 5s *(default)* $0.4704 · 10s $0.9408 #### Kling O1 References Default: **$0.4704** — 5s * **duration:** 5s *(default)* $0.4704 · 10s $0.9408 #### Kling O3 Pro References Default: **$0.4704** — 1080p · 5s | Duration \ Resolution | 1080p *(default)* | 4k | | --------------------- | ----------------- | ----- | | 3s | $0.28 | $1.58 | | 4s | $0.38 | $1.67 | | 5s *(default)* | $0.47 | $1.76 | | 6s | $0.56 | $1.86 | | 7s | $0.66 | $1.95 | | 8s | $0.75 | $2.05 | | 9s | $0.85 | $2.14 | | 10s | $0.94 | $2.23 | | 11s | $1.03 | $2.33 | | 12s | $1.13 | $2.42 | | 13s | $1.22 | $2.52 | | 14s | $1.32 | $2.61 | | 15s | $1.41 | $2.70 | #### Kling O3 Standard Default: **$0.4704** — 5s * **duration:** 3s $0.2822 · 4s $0.3763 · 5s *(default)* $0.4704 · 6s $0.5645 · 7s $0.6586 · 8s $0.7526 · 9s $0.8467 · 10s $0.9408 · 11s $1.0349 · 12s $1.1290 · 13s $1.2230 · 14s $1.3171 · 15s $1.4112 #### WAN 2.2 Animate Move Default: **$0.4800** — 720p * **resolution:** 480p $0.2400 · 580p $0.3600 · 720p *(default)* $0.4800 #### WAN 2.2 Animate Replace Default: **$0.4800** — 720p * **resolution:** 480p $0.2400 · 580p $0.3600 · 720p *(default)* $0.4800 #### Tencent Hunyuan Default: **$0.4807** — pro mode off * **pro mode:** on $0.9308 · off *(default)* $0.4807 #### Veo 3.1 Frames Lite Default: **$0.4807** — 720p · 8s | Duration \ Resolution | 720p *(default)* | 4k | | --------------------- | ---------------- | ----- | | 4s | $0.24 | $0.88 | | 6s | $0.36 | $1.00 | | 8s *(default)* | $0.48 | $1.12 | #### Veo 3.1 Lite Default: **$0.4807** — 8s * **duration:** 4s $0.2407 · 6s $0.3607 · 8s *(default)* $0.4807 #### WAN 2.2 Default: **$0.4817** — 720p · 5s | Duration \ Resolution | 480p | 580p | 720p *(default)* | | --------------------- | ----- | ----- | ---------------- | | 5s *(default)* | $0.32 | $0.41 | $0.48 | | 8s | $0.57 | $0.74 | $0.87 | #### Kling 3.0 Standard Default: **$0.5292** — 5s * **duration:** 3s $0.3175 · 4s $0.4234 · 5s *(default)* $0.5292 · 6s $0.6350 · 7s $0.7409 · 8s $0.8467 · 9s $0.9526 · 10s $1.0584 · 11s $1.1642 · 12s $1.2701 · 13s $1.3759 · 14s $1.4818 · 15s $1.5876 #### Kling O3 Standard Reference Default: **$0.5292** — 5s * **duration:** 3s $0.3175 · 4s $0.4234 · 5s *(default)* $0.5292 · 6s $0.6350 · 7s $0.7409 · 8s $0.8467 · 9s $0.9526 · 10s $1.0584 · 11s $1.1642 · 12s $1.2701 · 13s $1.3759 · 14s $1.4818 · 15s $1.5876 #### Kling O3 Standard References Default: **$0.5292** — 5s * **duration:** 3s $0.3175 · 4s $0.4234 · 5s *(default)* $0.5292 · 6s $0.6350 · 7s $0.7409 · 8s $0.8467 · 9s $0.9526 · 10s $1.0584 · 11s $1.1642 · 12s $1.2701 · 13s $1.3759 · 14s $1.4818 · 15s $1.5876 #### Kling O3 Pro Default: **$0.5880** — 1080p · 5s | Duration \ Resolution | 1080p *(default)* | 4k | | --------------------- | ----------------- | ----- | | 3s | $0.35 | $1.97 | | 4s | $0.47 | $2.09 | | 5s *(default)* | $0.59 | $2.21 | | 6s | $0.71 | $2.32 | | 7s | $0.82 | $2.44 | | 8s | $0.94 | $2.56 | | 9s | $1.06 | $2.68 | | 10s | $1.18 | $2.79 | | 11s | $1.29 | $2.91 | | 12s | $1.41 | $3.03 | | 13s | $1.53 | $3.15 | | 14s | $1.65 | $3.26 | | 15s | $1.76 | $3.38 | #### Lucy Edit Pro Default: **$0.6000** — 720p * **resolution:** 720p *(default)* $0.6000 · 480p $0.3000 #### WAN 2.5 Default: **$0.6005** — 720P · 5s | Duration \ Resolution | 720P *(default)* | 1080P | | --------------------- | ---------------- | ----- | | 5s *(default)* | $0.60 | $0.90 | | 10s | $1.20 | $1.80 | #### WAN 2.6 Default: **$0.6005** — 720P · 5s | Duration \ Resolution | 720P *(default)* | 1080P | | --------------------- | ---------------- | ----- | | 5s *(default)* | $0.60 | $0.90 | | 10s | $1.20 | $1.80 | | 15s | $1.80 | $2.70 | #### WAN 2.7 Default: **$0.6005** — 720P · 5s | Duration \ Resolution | 720P *(default)* | 1080P | | --------------------- | ---------------- | ----- | | 5s *(default)* | $0.60 | $0.90 | | 10s | $1.20 | $1.80 | | 15s | $1.80 | $2.70 | #### Kling 3.0 Standard (Turbo) Default: **$0.6720** — 5s * **duration:** 3s $0.4032 · 4s $0.5376 · 5s *(default)* $0.6720 · 6s $0.8064 · 7s $0.9408 · 8s $1.0752 · 9s $1.2096 · 10s $1.3440 · 11s $1.4784 · 12s $1.6128 · 13s $1.7472 · 14s $1.8816 · 15s $2.0160 #### Kling 2.6 Pro Default: **$0.6762** — 5s * **duration:** 5s *(default)* $0.6762 · 10s $1.3524 #### Kling 3.0 Pro Default: **$0.7056** — 1080p · 5s | Duration \ Resolution | 1080p *(default)* | 4k | | --------------------- | ----------------- | ----- | | 3s | $0.42 | $1.83 | | 4s | $0.56 | $1.98 | | 5s *(default)* | $0.71 | $2.12 | | 6s | $0.85 | $2.26 | | 7s | $0.99 | $2.40 | | 8s | $1.13 | $2.54 | | 9s | $1.27 | $2.68 | | 10s | $1.41 | $2.82 | | 11s | $1.55 | $2.96 | | 12s | $1.69 | $3.10 | | 13s | $1.83 | $3.25 | | 14s | $1.98 | $3.39 | | 15s | $2.12 | $3.53 | #### Kling O1 Reference Default: **$0.7056** — 5s * **duration:** 5s *(default)* $0.7056 · 10s $1.4112 #### Kling O3 Pro Reference Default: **$0.7056** — 5s * **duration:** 3s $0.4234 · 4s $0.5645 · 5s *(default)* $0.7056 · 6s $0.8467 · 7s $0.9878 · 8s $1.1290 · 9s $1.2701 · 10s $1.4112 · 11s $1.5523 · 12s $1.6934 · 13s $1.8346 · 14s $1.9757 · 15s $2.1168 #### Runway Gen-4.5 Default: **$0.7202** — 5s * **duration:** 5s *(default)* $0.7202 · 8s $1.1524 · 10s $1.4405 #### Seedance 1.0 Pro Default: **$0.7445** — 1080p · 5s | Duration \ Resolution | 480p | 720p | 1080p *(default)* | | --------------------- | ----- | ----- | ----------------- | | 3s | $0.09 | $0.20 | $0.45 | | 4s | $0.12 | $0.27 | $0.60 | | 5s *(default)* | $0.15 | $0.34 | $0.74 | | 6s | $0.18 | $0.40 | $0.89 | | 7s | $0.21 | $0.47 | $1.04 | | 8s | $0.24 | $0.54 | $1.19 | | 9s | $0.27 | $0.60 | $1.34 | | 10s | $0.30 | $0.67 | $1.49 | | 11s | $0.33 | $0.74 | $1.64 | | 12s | $0.36 | $0.80 | $1.79 | #### Aurora Default: **$0.8400** — 720p * **resolution:** 480p $0.4200 · 720p *(default)* $0.8400 #### Happy Horse 1.0 Default: **$0.8400** — 720p · 5s | Duration \ Resolution | 720p *(default)* | 1080p | | --------------------- | ---------------- | ----- | | 3s | $0.50 | $1.34 | | 4s | $0.67 | $1.51 | | 5s *(default)* | $0.84 | $1.68 | | 6s | $1.01 | $1.85 | | 7s | $1.18 | $2.02 | | 8s | $1.34 | $2.18 | | 9s | $1.51 | $2.35 | | 10s | $1.68 | $2.52 | | 11s | $1.85 | $2.69 | | 12s | $2.02 | $2.86 | | 13s | $2.18 | $3.02 | | 14s | $2.35 | $3.19 | | 15s | $2.52 | $3.36 | #### Happy Horse 1.0 Edit Default: **$0.8400** — 720p * **resolution:** 720p *(default)* $0.8400 · 1080p $1.6800 #### Happy Horse 1.0 References Default: **$0.8400** — 720p · 5s | Duration \ Resolution | 720p *(default)* | 1080p | | --------------------- | ---------------- | ----- | | 3s | $0.50 | $1.34 | | 4s | $0.67 | $1.51 | | 5s *(default)* | $0.84 | $1.68 | | 6s | $1.01 | $1.85 | | 7s | $1.18 | $2.02 | | 8s | $1.34 | $2.18 | | 9s | $1.51 | $2.35 | | 10s | $1.68 | $2.52 | | 11s | $1.85 | $2.69 | | 12s | $2.02 | $2.86 | | 13s | $2.18 | $3.02 | | 14s | $2.35 | $3.19 | | 15s | $2.52 | $3.36 | #### Kling 3.0 Pro (Turbo) Default: **$0.8400** — 5s * **duration:** 3s $0.5040 · 4s $0.6720 · 5s *(default)* $0.8400 · 6s $1.0080 · 7s $1.1760 · 8s $1.3440 · 9s $1.5120 · 10s $1.6800 · 11s $1.8480 · 12s $2.0160 · 13s $2.1840 · 14s $2.3520 · 15s $2.5200 #### Seedance 2.0 Fast Default: **$0.8514** — 720p · 5s | Duration \ Resolution | 480p | 720p *(default)* | | --------------------- | ----- | ---------------- | | 4s | $0.22 | $0.68 | | 5s *(default)* | $0.39 | $0.85 | | 6s | $0.56 | $1.02 | | 7s | $0.73 | $1.19 | | 8s | $0.90 | $1.36 | | 9s | $1.07 | $1.53 | | 10s | $1.24 | $1.70 | | 11s | $1.41 | $1.87 | | 12s | $1.58 | $2.04 | | 13s | $1.75 | $2.21 | | 14s | $1.92 | $2.38 | | 15s | $2.09 | $2.55 | #### Seedance 2.0 Reference (Fast) Default: **$0.8514** — 720p · 5s | Duration \ Resolution | 480p | 720p *(default)* | | --------------------- | ----- | ---------------- | | 4s | $0.22 | $0.68 | | 5s *(default)* | $0.39 | $0.85 | | 6s | $0.56 | $1.02 | | 7s | $0.73 | $1.19 | | 8s | $0.90 | $1.36 | | 9s | $1.07 | $1.53 | | 10s | $1.24 | $1.70 | | 11s | $1.41 | $1.87 | | 12s | $1.58 | $2.04 | | 13s | $1.75 | $2.21 | | 14s | $1.92 | $2.38 | | 15s | $2.09 | $2.55 | #### Luma Ray 2 Default: **$0.8525** — 720p · 5s · 16:9 | Duration \ Resolution | 540p | 720p *(default)* | | --------------------- | ----- | ---------------- | | 5s *(default)* | $0.48 | $0.85 | | 9s | $0.86 | $1.53 | * **aspect ratio:** 21:9 $1.1189 · 16:9 *(default)* $0.8525 · 4:3 $0.6394 · 1:1 $0.4795 · 3:4 $0.6394 · 9:16 $0.8525 · 9:21 $1.1189 #### Fabric 1.0 Default: **$0.9000** — 720p * **resolution:** 480p $0.4950 · 720p *(default)* $0.9000 #### Seedance 2.0 Mini Default: **$0.9288** — 720p · 5s | Duration \ Resolution | 480p | 720p *(default)* | | --------------------- | ----- | ---------------- | | 4s | $0.22 | $0.72 | | 5s *(default)* | $0.43 | $0.93 | | 6s | $0.64 | $1.14 | | 7s | $0.85 | $1.35 | | 8s | $1.06 | $1.56 | | 9s | $1.27 | $1.78 | | 10s | $1.49 | $1.99 | | 11s | $1.70 | $2.20 | | 12s | $1.91 | $2.41 | | 13s | $2.12 | $2.62 | | 14s | $2.33 | $2.83 | | 15s | $2.54 | $3.05 | #### Seedance 2.0 Default: **$1.0584** — 720p · 5s | Duration \ Resolution | 480p | 720p *(default)* | 1080p | 4k | | --------------------- | ----- | ---------------- | ----- | ----- | | 4s | $0.28 | $0.85 | $2.17 | $5.08 | | 5s *(default)* | $0.49 | $1.06 | $2.38 | $5.29 | | 6s | $0.70 | $1.27 | $2.59 | $5.50 | | 7s | $0.91 | $1.48 | $2.80 | $5.72 | | 8s | $1.12 | $1.69 | $3.02 | $5.93 | | 9s | $1.33 | $1.91 | $3.23 | $6.14 | | 10s | $1.55 | $2.12 | $3.44 | $6.35 | | 11s | $1.76 | $2.33 | $3.65 | $6.56 | | 12s | $1.97 | $2.54 | $3.86 | $6.77 | | 13s | $2.18 | $2.75 | $4.07 | $6.99 | | 14s | $2.39 | $2.96 | $4.29 | $7.20 | | 15s | $2.60 | $3.18 | $4.50 | $7.41 | #### Seedance 2.0 Reference Default: **$1.0584** — 720p · 5s | Duration \ Resolution | 480p | 720p *(default)* | 1080p | 4k | | --------------------- | ----- | ---------------- | ----- | ----- | | 4s | $0.28 | $0.85 | $2.17 | $5.08 | | 5s *(default)* | $0.49 | $1.06 | $2.38 | $5.29 | | 6s | $0.70 | $1.27 | $2.59 | $5.50 | | 7s | $0.91 | $1.48 | $2.80 | $5.72 | | 8s | $1.12 | $1.69 | $3.02 | $5.93 | | 9s | $1.33 | $1.91 | $3.23 | $6.14 | | 10s | $1.55 | $2.12 | $3.44 | $6.35 | | 11s | $1.76 | $2.33 | $3.65 | $6.56 | | 12s | $1.97 | $2.54 | $3.86 | $6.77 | | 13s | $2.18 | $2.75 | $4.07 | $6.99 | | 14s | $2.39 | $2.96 | $4.29 | $7.20 | | 15s | $2.60 | $3.18 | $4.50 | $7.41 | #### Veo 2 Default: **$1.0802** — 5s * **duration:** 5s *(default)* $1.0802 · 6s $1.3395 · 7s $1.5979 · 8s $1.8571 #### Grok Imagine Default: **$1.3036** — 6s * **duration:** 1s $0.2143 · 2s $0.4321 · 3s $0.6500 · 4s $0.8679 · 5s $1.0857 · 6s *(default)* $1.3036 · 7s $1.5214 · 8s $1.7393 · 9s $1.9571 · 10s $2.1750 · 11s $2.3928 · 12s $2.6107 · 13s $2.8285 · 14s $3.0464 · 15s $3.2643 #### Grok Imagine References Default: **$1.3036** — 6s * **duration:** 1s $0.2143 · 2s $0.4321 · 3s $0.6500 · 4s $0.8679 · 5s $1.0857 · 6s *(default)* $1.3036 · 7s $1.5214 · 8s $1.7393 · 9s $1.9571 · 10s $2.1750 · 11s $2.3928 · 12s $2.6107 · 13s $2.8285 · 14s $3.0464 · 15s $3.2643 #### Kling 2.1 Master Default: **$1.3810** — 5s * **duration:** 5s *(default)* $1.3810 · 10s $2.7619 #### Sora 2 Pro Default: **$1.4400** — 720p · 4s | Duration \ Resolution | 720p *(default)* | 1080p | | --------------------- | ---------------- | ----- | | 4s *(default)* | $1.44 | $2.39 | | 8s | $2.88 | $4.78 | | 12s | $4.32 | $7.17 | #### Veo 3.1 (Fast) Default: **$1.4400** — 720p · 8s | Duration \ Resolution | 720p *(default)* | 4k | | --------------------- | ---------------- | ----- | | 4s | $0.96 | $2.87 | | 6s | $1.20 | $3.11 | | 8s *(default)* | $1.44 | $3.36 | #### Veo 3.1 Frames (Fast) Default: **$1.4400** — 720p · 8s | Duration \ Resolution | 720p *(default)* | 4k | | --------------------- | ---------------- | ----- | | 4s | $0.96 | $2.87 | | 6s | $1.20 | $3.11 | | 8s *(default)* | $1.44 | $3.36 | #### Marey Default: **$1.8005** — 5s * **duration:** 5s *(default)* $1.8005 · 10s $3.6010 #### Seedance 2.5 Default: **$2.4846** — 720p · 5s · 16:9 | Duration \ Resolution | 480p | 720p *(default)* | | --------------------- | ------ | ---------------- | | 4s | $0.66 | $1.99 | | 5s *(default)* | $1.16 | $2.48 | | 6s | $1.65 | $2.98 | | 7s | $2.15 | $3.48 | | 8s | $2.65 | $3.98 | | 9s | $3.15 | $4.47 | | 10s | $3.64 | $4.97 | | 11s | $4.14 | $5.47 | | 12s | $4.64 | $5.96 | | 13s | $5.13 | $6.46 | | 14s | $5.63 | $6.96 | | 15s | $6.13 | $7.45 | | 16s | $6.62 | $7.95 | | 17s | $7.12 | $8.45 | | 18s | $7.62 | $8.94 | | 19s | $8.11 | $9.44 | | 20s | $8.61 | $9.94 | | 21s | $9.11 | $10.44 | | 22s | $9.61 | $10.93 | | 23s | $10.10 | $11.43 | | 24s | $10.60 | $11.93 | | 25s | $11.10 | $12.42 | | 26s | $11.59 | $12.92 | | 27s | $12.09 | $13.42 | | 28s | $12.59 | $13.91 | | 29s | $13.08 | $14.41 | | 30s | $13.58 | $14.91 | * **aspect ratio:** 16:9 *(default)* $2.4846 · 9:16 $2.4846 · 4:3 $1.8635 · 3:4 $1.8635 · 1:1 $1.3976 · 21:9 $3.2611 #### Veo 3.1 Default: **$3.8400** — 720p · 8s | Duration \ Resolution | 720p *(default)* | 4k | | --------------------- | ---------------- | ----- | | 4s | $3.36 | $5.28 | | 6s | $3.60 | $5.52 | | 8s *(default)* | $3.84 | $5.76 | #### Veo 3.1 Frames Default: **$3.8400** — 720p · 8s | Duration \ Resolution | 720p *(default)* | 4k | | --------------------- | ---------------- | ----- | | 4s | $3.36 | $5.28 | | 6s | $3.60 | $5.52 | | 8s *(default)* | $3.84 | $5.76 | #### Veo 3.1 Ingredients Default: **$3.8400** — 720p * **resolution:** 720p *(default)* $3.8400 · 1080p $3.8400 · 4k $5.7600 ## Image Models #### Qwen Image Edit Default: **$0.0360** — auto * **aspect ratio:** auto *(default)* $0.0360 · 21:9 $0.0154 · 16:9 $0.0203 · 4:3 $0.0270 · 3:2 $0.0239 · 1:1 $0.0360 · 2:3 $0.0239 · 3:4 $0.0270 · 9:16 $0.0203 · 9:21 $0.0154 #### Qwen Image Edit Plus Default: **$0.0360** — auto * **aspect ratio:** auto *(default)* $0.0360 · 21:9 $0.0154 · 16:9 $0.0203 · 4:3 $0.0270 · 3:2 $0.0239 · 1:1 $0.0360 · 2:3 $0.0239 · 3:4 $0.0270 · 9:16 $0.0203 · 9:21 $0.0154 #### Qwen Image 2.0 Default: **$0.0420** — auto * **aspect ratio:** auto *(default)* $0.0420 · 21:9 $0.0180 · 16:9 $0.0236 · 4:3 $0.0315 · 3:2 $0.0279 · 1:1 $0.0420 · 2:3 $0.0279 · 3:4 $0.0315 · 9:16 $0.0236 · 9:21 $0.0180 #### Grok Imagine Quality Default: **$0.0600** — 1k * **resolution:** 1k *(default)* $0.0600 · 2k $0.3000 #### Ideogram 4.0 Default: **$0.0720** — rendering speed DEFAULT * **rendering speed:** TURBO $0.0360 · DEFAULT *(default)* $0.0720 · QUALITY $0.1200 #### Nano Banana 2 Default: **$0.0720** — 1K · web search off * **resolution:** 1K *(default)* $0.0720 · 2K $0.1080 · 4K $0.1440 * **web search:** on $0.0900 · off *(default)* $0.0720 #### Seedream 5 Pro Default: **$0.1699** — 2k * **resolution:** 1k $0.0850 · 2k *(default)* $0.1699 #### Enhancor V1 Default: **$0.1206** — enhancement standard * **enhancement:** standard *(default)* $0.1206 · heavy $0.3304 #### Magnific Creative Upscaler Default: **$0.1206** — scale factor 2x * **scale factor:** 2x *(default)* $0.1206 · 4x $0.2412 · 8x $0.4824 · 16x $0.9648 #### Riverflow 2.0 Pro Default: **$0.1620** — 2K * **resolution:** 1K $0.1620 · 2K *(default)* $0.1620 · 4K $0.3240 #### Riverflow 2.0 Pro Inpainting Default: **$0.1620** — 2K * **resolution:** 1K $0.1620 · 2K *(default)* $0.1620 · 4K $0.3240 #### Nano Banana Pro Default: **$0.1800** — 1K * **resolution:** 1K *(default)* $0.1800 · 2K $0.1800 · 4K $0.3600 #### Nano Banana Pro Inpainting Default: **$0.1800** — 1K * **resolution:** 1K *(default)* $0.1800 · 2K $0.1800 · 4K $0.3600 #### GPT Image 2 Default: **$0.2635** — 1k · quality high * **resolution:** 1k *(default)* $0.2635 · 2k $0.5196 · 4k $0.8741 * **quality:** low $0.0174 · medium $0.0736 · high *(default)* $0.2635 #### Riverflow 2.5 Pro Default: **$0.2832** — 2K · thinkingLevel medium * **resolution:** 1K $0.2832 · 2K *(default)* $0.2832 · 4K $0.5664 * **thinkingLevel:** low $0.2712 · medium *(default)* $0.2832 · high $0.3912 · xhigh $0.5856 #### Enhancor V4 Default: **$0.3367** — fast mode on * **fast mode:** off $0.9597 · on *(default)* $0.3367 #### Enhancor V3 Default: **$0.4663** — 1024 * **output res:** 1024 *(default)* $0.4663 · 2048 $0.6041 · 3072 $0.8372 # Launch Bonus {% hint style="success" %} **Extended through August 31, 2026.** Every paid plan gets extra usage on top of its baseline. {% endhint %} Through the launch bonus period, your plan stretches further. Existing usage rules continue to apply for everything outside the bonus. ## Bonus Usage on Every Plan Each paid plan includes extra usage in addition to its standard monthly budget. The bonus is added to your workspace automatically and resets with your billing cycle, just like your standard budget. | Plan | Standard usage | Launch bonus | Total during launch | | ------- | -------------- | ------------ | ------------------- | | Starter | $18/seat | +$12/seat | $30/seat | | Pro | $50/seat | +$50/seat | $100/seat | | Max | $200/seat | +$100/seat | $300/seat | Free, Enterprise, and custom plans aren't part of the launch bonus. ## How Many Generations Per Plan How far does each plan's launch-bonus budget go? The numbers below show roughly how many generations of each model fit within the bonus-period budget — Starter $30, Pro $100, Max $300 — if the entire budget went to that one model. Real workflows mix models, so use these as a rough scale rather than a target. ### Image | Model | Starter ($30) | Pro ($100) | Max ($300) | | -------------------- | ------------- | ---------- | ---------- | | Flux 2 | \~3,900 | \~13,200 | \~39,000 | | Seedream 4.0 | \~1,200 | \~4,000 | \~11,900 | | Flux 2 Pro | \~830 | \~2,800 | \~8,300 | | Nano Banana | \~800 | \~2,700 | \~7,900 | | Stable Diffusion 3.5 | \~710 | \~2,400 | \~7,000 | | Recraft V4 | \~620 | \~2,000 | \~6,200 | | Imagen 4 | \~500 | \~1,700 | \~5,000 | | Ideogram 3.0 | \~420 | \~1,400 | \~4,200 | | Nano Banana 2 | \~420 | \~1,400 | \~4,200 | | Nano Banana Pro | \~210 | \~700 | \~2,100 | | GPT Image 1.5 | \~180 | \~620 | \~1,900 | ### Video | Model | Starter ($30) | Pro ($100) | Max ($300) | | ------------------- | ------------- | ---------- | ---------- | | Minimax Hailuo | \~230 | \~730 | \~2,200 | | Kling 2.5 Turbo Pro | \~100 | \~340 | \~1,000 | | LTX-2 Pro | 69 | \~230 | \~690 | | Veo 3.1 Lite | 69 | \~230 | \~690 | | Pika | 56 | \~190 | \~560 | | Kling 2.6 Pro | 50 | \~170 | \~510 | | Luma Ray 2 | 35 | \~120 | \~350 | | Sora 2 Pro | 20 | 69 | \~210 | | Seedance 2.0 | 17 | 55 | \~170 | | Veo 3.1 | 8 | 28 | 86 | ### Text | Model | Starter ($30) | Pro ($100) | Max ($300) | | ----------------- | ------------- | ---------- | ---------- | | Claude Sonnet 4.6 | \~4,200 | \~13,900 | \~41,700 | | Claude Opus 4.7 | \~3,300 | \~11,100 | \~33,300 | | Gemini 3.1 Pro | \~1,200 | \~4,000 | \~11,900 | | GPT-5.2 | \~930 | \~3,100 | \~9,300 | | GPT-5.4 | \~660 | \~2,200 | \~6,700 | | o3 Deep Research | 36 | \~120 | \~370 | ### Audio | Model | Starter ($30) | Pro ($100) | Max ($300) | | -------------------------- | ------------- | ---------- | ---------- | | ElevenLabs Multilingual v2 | \~240 | \~830 | \~2,500 | | ElevenLabs Sound Effects | \~210 | \~700 | \~2,100 | | ElevenLabs Music v1 | 83 | \~280 | \~830 | Counts assume an even average cost per generation; text generations especially vary by length. For published per-generation rates, see the [Model Pricing](/plans-and-billing/model-pricing) page. ## What Changes When the Launch Bonus Ends When the launch bonus ends, bonus usage rolls off and your plan returns to its standard monthly budget. ## How It Stacks With Teams The bonus usage scales per seat. A 4-person Pro workspace, for example: * Standard shared budget: 4 × $50 = **$200/month** * Launch bonus: 4 × $50 = **+$200/month** (during launch) * Total shared budget during launch: **$400/month** Per-member caps and pooled-vs-per-user budget toggles work the same way during launch as they do normally. *** ## Questions? * General: * Enterprise: [flora.ai/contact](https://flora.ai/contact) * For how the standing pricing works once the launch bonus ends, see [How FLORA Pricing Works](/plans-and-billing/pricing). # Pricing FAQ {% hint style="success" %} **Launch bonus extended through August 31, 2026.** Every paid plan gets extra usage during launch. See the [Launch Bonus](/plans-and-billing/launch-bonus) page. {% endhint %} ## Launch Bonus ### What's the launch bonus? Every paid plan gets extra usage on top of its standard monthly budget through August 31, 2026 — Starter +$12, Pro +$50, Max +$100 per seat. See [Launch Bonus](/plans-and-billing/launch-bonus) for full details and generation estimates. ### What changes when the launch bonus ends? Plans return to their published baselines: Starter $18 usage, Pro $50 usage, Max $200 usage per seat — no change to subscription price. ## Legacy Contributors ### What's a legacy contributor? If you admin a **multi-seat usage-based workspace** that migrated in May 2026, your non-admin teammates may still be **legacy contributors** — a temporary role during the post-migration grace period. They keep full editing access and pull from your workspace's pooled usage, so day-to-day work doesn't change. The [launch bonus](/plans-and-billing/launch-bonus) applies to paid seats only — to include a teammate, convert them to a paid seat. **If you were on Pro per-seat, Agency Standard, or Agency Elite**, you don't have legacy contributors — your team migrated directly to per-seat billing. The grace period runs **six months from migration** (until **November 5, 2026** for monthly workspaces; annual workspaces until their first renewal after May 5, 2026). After it ends, legacy contributors lose editing access until you buy a paid seat for them. ### How do I add a paid seat? You can buy up to 8 paid seats through self-serve billing. Past 8, the standard path is moving to Enterprise — though in some cases we can raise the cap on a per-workspace basis (email to ask). If your workspace has legacy contributors, the new seat first converts one of them to a permanent editor — see the [next question](#how-do-i-convert-a-legacy-contributor-to-a-paid-seat). See [Seats](/plans-and-billing/manage-subscription#seats) for the full picture. ### How do I convert a legacy contributor to a paid seat? Each new seat you buy converts one of your legacy contributors into a permanent editor — you pick which one. You can't assign a fresh seat to a brand-new person until all your legacy contributors have been converted (or removed). See [Manage Your Subscription → Seats](/plans-and-billing/manage-subscription#seats) for the full picture. ### What is seat cycling? Seat cycling is how you bring on a new editor without adding a paid seat — useful if you're at the seat cap or just don't want to grow your subscription. Demote an existing editor to **guest** (which frees their seat without changing your subscription cost), then assign the freed seat to the new editor. You can promote a guest back to editor later if you want to swap again. ## Usage & Billing ### How is usage calculated? Each generation has a dollar cost based on the model used, which is deducted from your workspace's monthly budget. ### What counts as usage? Any generation from an AI model — images, video, audio, text — based on that model's published cost. ### Do failed generations count against my usage? No. If a generation fails, nothing is deducted. ### Where do I see what each model costs? Three places: * [**Model Pricing**](/plans-and-billing/model-pricing) page — published dollar rates per model, for comparing before you generate. * **Usage History** view — per-generation dollar cost and percentage of budget consumed. * **App header** — live percentage meter showing budget remaining at a glance. ### Can I see exact dollar costs for each generation? Yes. The **Usage History** view shows cost per generation, per project, and per member — useful for client billing or team budgets. ### Where do I see my current usage? A live percentage bar is in the app header. For the current cycle, open **Settings → People & Seats** and check **Usage this period** — included usage, your **reset date**, and any overage for the period. Per-generation detail is in **Usage History** under **Plan & Usage**. ### When does my usage reset? Your monthly included usage resets on your workspace billing date. The exact time appears under **People & Seats → Usage this period** (for example, *Resets on July 1st at 11 PM PDT*). Launch bonus usage resets on the same schedule as your standard monthly budget. ### Does my monthly usage roll over if I don't use it all? No. Your plan's monthly budget resets each cycle — but any extra usage you have (from migrated credits, referrals, promotions, or support grants) persists. ### What's the difference between monthly and annual billing? Annual saves 20% on the subscription price — your monthly usage budget is the same either way. ## Plans & Models ### Can I change plans anytime? Yes. Upgrades take effect immediately. Downgrades apply at the end of your current billing cycle. ### Do all plans have access to every model? **In the FLORA app:** **Starter and above** include **text, image, and video** models (plus the rest of the catalog, e.g. audio), all at the same published per-generation rates. **Free** includes **text and image models only** — no video models in the canvas — with **limited complimentary generations**, not a monthly dollar budget. See [How Pricing Works](/plans-and-billing/pricing) and [flora.ai/pricing](https://flora.ai/pricing). **FLORA API and MCP:** Available on **Starter and above**, not on **Free**. Upgrade to Starter for programmatic access and agent integrations. Some workspace and team features vary by plan — see [flora.ai/pricing](https://flora.ai/pricing) for the full feature comparison. ### Is FAUNA really free on every plan? Yes. FAUNA messages and interactions don't count against your usage budget on any plan, ever. ### Does the Free plan have any other limits? Yes: * **Projects** — Free is capped at 3 active projects. All paid plans have unlimited projects. * **Models** — Free is **text and image** only in the app (no video models). Starter and above add video and the full catalog at published rates. * **Explore usage** — Free uses **limited complimentary generations**, not a monthly dollar pool. Paid plans include a monthly usage budget. See [How Pricing Works](/plans-and-billing/pricing) and [flora.ai/pricing](https://flora.ai/pricing) for the current free cap. * **API and MCP** — Not included on Free; included from **Starter** upward. ### Can I buy extra usage outside my plan? Not as a direct purchase. To go beyond your monthly budget, enable on-demand spending or upgrade your plan. ### What's on-demand spending? An opt-in way to keep generating after your monthly budget runs out. Workspace admins enable it under **Settings → People & Seats** (shown as **Overage**). You set a **max overage** dollar limit for the billing period so spending cannot run away. A payment method must be on file on self-serve plans. Same published model rates as included usage — no markup. ### When am I charged for on-demand / overage usage? **Not at the end of the billing cycle.** On self-serve plans (Starter, Pro, Max), your card is charged when unpaid overage crosses billing thresholds — often **while you are still generating**, not on renewal day. Typical order: 1. **You hit your max overage cap** — total overage for the period reaches the limit you set; Flora bills any remaining unpaid amount and stops further overage until renewal. 2. **Unpaid overage reaches $200** — system safety threshold. 3. **Unpaid overage reaches your workspace's monthly subscription total** — plan price × editor seats (monthly equivalent on annual plans). Example: solo Starter ($18/mo) is often charged around **$18** of unpaid overage at a time; a 4-seat Pro workspace ($200/mo) around **$200** at a time. 4. **Daily balance check** — if more than **$5** is still unpaid and none of the above applied yet, Flora tries to collect it once per day (around 8:15 AM UTC). You may see **several charges in one period** if your max overage is higher than your subscription total — each charge clears unpaid overage, but your **period total** still counts toward your cap. **Enterprise** workspaces on invoice billing are usually invoiced on a **monthly or quarterly** contract schedule instead of charged to a card immediately. See **People & Seats** for the next invoice date when that applies. ### What does the max overage cap do? Two things: * **Spending limit** — Flora blocks new generations when **total overage this period** would exceed your cap. * **Billing trigger** — when total overage **reaches** your cap, Flora bills whatever is still unpaid and overage stops for the rest of the period. Paying an earlier overage charge does **not** give you a fresh cap mid-cycle — the period total keeps accumulating until subscription renewal. If your cap is **$10** on a **$18/mo** Starter workspace, you will be billed and stopped around **$10**, not $18. If your cap is **$100**, you may be charged repeatedly near **$18** unpaid until you reach **$100** total for the period. ### Is on-demand usage priced the same as included usage? Yes. Same published rates, no markup. ### What happens if I cancel? You keep your plan's access through the end of the billing period. Any extra usage in your workspace stays with you and doesn't expire if you come back. ### Do you offer student or education discounts? Yes. Learn more at [flora.ai/edu](https://flora.ai/edu). ## Teams & Workspaces ### How do shared workspaces work? Every paid seat adds to a shared usage pool, so your whole team draws from the same budget. ### Can I set a spending cap for individual team members? Yes. Workspace admins can set per-member usage caps. ### Can admins split usage per-user instead of pooling it? Yes. Admins can toggle between a pooled budget and per-user budgets at any time. ### Can I use FLORA with collaborators who aren't on my workspace? Yes. They can join a shared project and generate using their own workspace's usage pool. ### What if I need more than 8 seats? Reach out at [flora.ai/contact](https://flora.ai/contact). *** ## Questions? * General: * Enterprise: [flora.ai/contact](https://flora.ai/contact) --- [Next Page](/llms-full.txt/1)