For the complete documentation index, see llms.txt. This page is also available as Markdown.

Batch Generate

Build a table of prompts and generate images in bulk

Batch Generate is a spreadsheet for image generation. Each row holds one prompt and its settings; press Generate and every row runs at once. It's the fastest way to produce a hundred product shots, try one idea across twenty styles, or work through a list you already have in a spreadsheet—without wiring up a single node.

Open it from Batch Generate in the left sidebar of your dashboard.

Batch Generate is a full page of its own, separate from the canvas. If you want bulk processing inside a canvas workflow—feeding many items through nodes you've already connected—use the Batch Node instead.

Overview

  • One row, one image — write a prompt, pick a model, and the result appears in the same row

  • Fill a table in seconds — expand a single prompt template into dozens of rows automatically

  • Change many rows at once — set the model, resolution, or aspect ratio across the whole batch from the column header

  • Bring your own images — attach reference images to the whole batch or to individual rows

  • Generate everything in one click — with a cost estimate shown before anything runs

  • Review in place — filter by status, search prompts, switch between a table and a grid of thumbnails, and download results individually or as a zip

  • Nothing to save — batches save themselves and stay in your workspace


Getting Started

1

Open Batch Generate

Click Batch Generate in the left sidebar. A new, empty batch is created for you with one blank row.

2

Write a prompt

Click into the Prompt cell of the first row and describe the image you want.

3

Choose a model and settings

Pick a model in the Model column, and optionally set a Resolution and Aspect ratio. Leaving them on Default and Auto lets the model decide.

4

Add more rows

Use Add row for one more empty row, or Add in bulk to create many rows at once from a template.

5

Generate

The bar at the bottom shows how many rows are ready—click Generate, confirm the cost, and every row runs. Results fill in the Output column as they finish.


Building Your Batch

The columns

Column
What it holds

#

The row number

Prompt

What you want generated—the only field you always have to fill in

Input

Reference images for this row (optional)

Model

Which image model runs this row

Resolution

Output size, if the model offers a choice

Aspect ratio

Output shape, if the model offers a choice

Output

The generated image once the row has run

Adding rows one at a time

Add row appears above an empty batch and in the footer below the table. Each click adds one blank row using the batch's current model.

Adding rows in bulk

Add in bulk is where Batch Generate earns its name. Instead of typing forty similar prompts, you write one prompt template, mark the parts that change, and list the options for each.

1

Write the template

Click Add in bulk and describe the image, leaving out the details that vary: "a product photo of a ceramic mug on a background, studio lighting".

2

Insert variables

Put your cursor where a detail should change and insert a text variable. Name it something obvious—color, background, season—and it appears in the prompt as a colored pill. Add as many as you need.

3

Fill in the values

Each variable gets its own box below the template. Type a value and press Enter to add it; repeat for every option you want. So color might get matte black, sage green, and cream, while background gets marble and linen.

4

Check the preview

FLORA writes out every combination—three colors × two backgrounds = six rows—and lists them in a preview with a running row count. Adjust the values until the count looks right.

5

Add the rows

Click Add rows and they drop straight into the table, ready to generate.

No variables? Use Copies. If your template has no variables, a Copies field appears instead—set it to 10 and you'll get ten identical rows. Useful when you want several takes on the same prompt.

Image variables

Variables can hold images as well as text. Insert an image variable, upload one or more images as its values, and FLORA creates a row per image—attaching that image to the row and referencing it inline in the prompt. This is how you run the same treatment across a folder of products, garments, or headshots.

Changing many rows at once

The Model, Resolution, and Aspect ratio column headers are menus. Pick a value and it applies to every row in the batch—or only to the rows you've selected, if you have a selection. Changing the model on a whole batch is one click, not one hundred.

Duplicating and deleting rows

Hover a row and open its menu for Regenerate, Duplicate, and Delete. Duplicate is handy when you want to keep a prompt and tweak one word.

To remove several rows at once, tick their checkboxes and use Delete in the toolbar that appears.


Adding Input Images

Some models can take reference images alongside the prompt—a product to keep consistent, a style to match, a person to place in a scene. Batch Generate gives you two places to attach them.

Images for the whole batch

The Input images tray above the table holds references shared by every row. Upload a logo, a colour card, or a product shot here and it's attached to all rows automatically—no need to repeat yourself.

Images for one row

Click the Input cell on any row to add references to just that row. Use this when each row has its own subject.

Mentioning an image in a prompt

Once an image is attached to a row, you can point at it from inside the prompt: type @ and pick the image. This tells the model which reference you mean—"put @product on a marble surface"—rather than leaving it to guess.

An @ mention only works for an image that's actually attached to that row. If you remove the image, the row will ask you to clear the mention before it can generate.


Models and Settings

Batch Generate runs image models. Pick one per row—rows in the same batch can use different models, which makes it easy to compare them side by side on identical prompts.

Model
Provider
Notes

Nano Banana 2

Google

Google's state-of-the-art image model — the default for new batches

Nano Banana Pro

Google

Google's highest quality, up to 4K

GPT Image 2

OpenAI

OpenAI's flagship image model

Seedream 5.0

Bytedance

Next-generation image synthesis with enhanced prompt understanding

Krea 2 Large

Krea

High-fidelity, stylized image model

The model list reflects what's enabled for your workspace, so it may be shorter than the list above. Resolution and Aspect ratio show only the options the selected model supports; leave them on Default and Auto to accept the model's own choice. See Model Pricing for what each one costs per image.


Generating

Running the batch

When at least one row is ready, a bar appears at the bottom of the page: "12 of 20 rows will generate." Click Generate 12 and FLORA confirms the estimated cost before charging anything.

Only rows without an output are included, so you can keep adding rows to a batch and generate the new ones without paying to redo the old ones.

Retrying and regenerating

  • Retry failed — if some rows fail, the bar offers to rerun just those

  • Regenerate a row — use the row's menu, or the button on a result card in grid view, to run it again with its current settings

  • Regenerate a selection — tick several rows and use Regenerate in the toolbar

If a row can't run

A row will flag itself rather than fail silently. The usual reasons:

  • An input image is unavailable — remove it from the row and try again

  • The model doesn't support that many input images — reduce them, or switch to a model that does

  • The model is unavailable — choose another one

  • The prompt is too long — prompts are capped at 8,000 characters


Reviewing Results

Two views

Use the toggle in the toolbar to switch between:

  • Row view — the full table, with each result as a thumbnail in its own row next to the prompt that made it

  • Grid view — just the images, as a wall of thumbnails, for judging results quickly

Click any result to open it full screen.

Finding rows

  • Status tabs — filter to All, Draft (not generated yet), Generating, Completed, or Failed. Each tab shows a count.

  • Search — type into the search box to narrow the table to rows whose prompts match.

Downloading

  • Download all downloads every completed result in the batch

  • Tick specific rows and use Download in the toolbar for just those

  • Download a single result from its row menu or its card in grid view

One image downloads as a file; several download together as a zip.

Downloading exports requires a paid plan, and bulk downloads require a plan that includes bulk export. See How Pricing Works for what's included in each plan.


Managing Batches

Batches save themselves as you work—there's no save button.

  • Rename — double-click the batch name at the top of the page, or use Rename this batch in the menu beside it

  • Switch batchesPrevious batches lists everything you've made, most recent first

  • Start a new batchNew batch opens a fresh table; the current one stays saved

  • DeleteDelete this batch removes the batch and all of its rows


Limits

Limit
Details

Rows per batch

200

Input images per row

4, including any shared batch input images

Prompt length

8,000 characters

Output type

Images only—Batch Generate doesn't produce video, audio, or text

Parallel generations

Free plans run generations one at a time; running a batch in parallel requires a paid plan


Tips

  • Prove the prompt on one row first. Generate a single row, check it, then bulk-expand once you're happy—cheaper than regenerating 200 rows.

  • Let variables do the typing. Anywhere you'd copy a prompt and change one word, use a text variable instead.

  • Compare models fairly. Duplicate a row a few times, change only the model on each copy, and generate—same prompt, same references, honest comparison.

  • Put shared references in the batch tray. Anything every row needs—a logo, a colour reference—belongs at the batch level, not pasted into each row.

  • Name your batches. "Spring lookbook — v2" is far easier to find in Previous batches than "Untitled".

  • Use the status tabs while it runs. Watch Failed during a big run and retry as you go instead of waiting for the whole batch.


Common Use Cases

  • Product catalogs — one row per SKU, with an image variable holding your product shots and a shared prompt for the scene

  • Colourway and variant exploration — a color variable and a material variable turn into every combination automatically

  • Campaign variations — the same key visual across a list of markets, seasons, or messages

  • Model bake-offs — one prompt, five models, results side by side in grid view

  • Working from a list you already have — paste values from a spreadsheet into a variable and let the batch expand

  • Volume drafts — set Copies to 20 and pick the best take

Last updated

Was this helpful?