Style Lab

An iterative testing environment for refining your style guide before generating images at scale.

Why Use the Style Lab?

Running a full batch to test a new style wastes credits if the style isn't right. The Style Lab lets you test with just 3 sample images per round, so you can iterate quickly and cheaply before committing.

Think of it as a sandbox — experiment freely, then "lock in" the style when you're satisfied.

How It Works

Edit Your System Prompt

Start with your current style guide or write a new one from scratch. The textarea shows the full prompt text.

Pick Test Items

The lab selects 3 random items from your queue as test subjects. Click Shuffle to get different items if you want to test with specific types.

Generate a Round

Click generate to create one image per test item with the current prompt. This costs 3 credits and takes about 45 seconds.

Review Results

See all 3 generated images side by side. Evaluate whether the style matches your vision.

Leave Feedback & Iterate

Describe what needs to change (e.g., "lighting is too harsh, need softer shadows"). Click Apply & Regenerate — the AI revises the prompt and automatically generates a new round with the updated style.

You can repeat steps 4-5 as many times as needed. Each round is tracked in the history so you can compare how the style evolved across iterations.

Locking In Your Style

When you're happy with the results, click Lock in Style. This saves the tested prompt as your project's Style Guide, so all future batch generations will use it.

The previous style guide is preserved in version history — you can always revert if needed.

Experiment History

Every round is saved with its prompt, test items, and generated images. The history page shows all rounds (newest first), letting you:

•Compare images across rounds to see how the style improved
•View the exact prompt used for each round
•Review the feedback that drove each revision
•Go back to a previous round's prompt if a later iteration went off track