Ready to analyze
IMAGE → PROMPT
Turn a reference image into a better prompt.
Upload a reference, add examples, and keep refining one precise prompt with the Assistant.
CHOOSE A WORKFLOW
What do you want to make?
MODEL SETTINGS
Choose the model for each workflow
Selections are saved for your next visit. Costs are estimates shown by the platform.
Vision models that can inspect reference images.
Text models for rewriting and refining prompts.
Reference-image editors that accept an image plus an instruction.
Model discovery is unavailable, so the built-in compatible defaults are being used.
01 / REFERENCE
Add a reference image
No image selected
Your image preview will appear here
01 / REFERENCE
Start with a prompt
The Assistant will keep your intent, improve the visual structure, and return a prompt, tags, negative prompt, and image specs.
02 / DIRECTION
Set the level of visual detail
Higher values ask for precise visual detail without inventing facts.
The Assistant returns the prompt, tags, negative prompt, and image specs together.
Higher values make only the requested edits.
Higher values explore the instruction while keeping characters and art style unless asked otherwise.
Upload a reference, describe the change, then balance preservation and invention.
This vision model inspects your reference for the next prompt.
03 / RESULTS
Your prompt is ready
Tags & weights
Negative prompt
Image specs
Generator and resolution are estimates unless embedded metadata is available.
04 / REFINE
Talk to the Assistant
Ask for an edit, addition, or correction. The current image and prompt stay in context.
03 / RESULTS
Your edited image is ready
Your edited image will appear here.
EDIT APPLIED