Batch Generate

Generate hundreds of images in a single run with real-time streaming progress.

Prerequisites

Before starting a batch run, you need:

•Items in your queue — Add items via Brainstorm chat
•A style guide — Set one in the Style Guide page, or a default will be used
•Sufficient credits — Cost varies by model selection (see Credits & Pricing)

The Generation Pipeline

For each item in your queue, three steps happen in sequence:

Prompt Writing

The AI reads your style guide and the item name, then writes a detailed text-to-image prompt optimized for your selected model. In Auto mode, it also classifies item complexity to pick the best model.

Image Generation

Your selected model renders the image from the prompt. Speed and cost depend on the model — from ~1 second with FLUX Schnell to ~15 seconds with FLUX Pro.

Background Removal

If enabled, BiRefNet v2 removes the background to create a transparent PNG. The original version is also saved when background removal is on.

During Generation

While a batch is running, the page shows:

•The item currently being generated
•A progress bar with percentage complete
•Success and failure counts
•Generated images appearing in real time as they complete
•Estimated time remaining

You can stop a batch at any time. Stopping finishes the current item but doesn't cancel images mid-generation. Items that haven't started yet remain in the queue for next time.

If you navigate away and come back while a batch is running, the page will automatically reconnect to the active run and show the current progress.

Cost

Component	Model	Cost
Prompt writing	Asset Forge Prompt Engine	Included
Image generation	Depends on model	10–40 credits
Background removal	BiRefNet v2	40 credits

Total varies by model: 10–80 credits per image depending on model and background removal setting. See Settings for the full model list.

Excluded Items

Items can be excluded from batch generation if the AI determines they're unsuitable (e.g., duplicates, inappropriate items). Excluded items are shown separately with the reason for exclusion. You can review and re-add them if the exclusion was incorrect.