**If you're happy with the result, please support me by pressing the like button 😄**

## What does this app do?
This AI image editor lets you transform your pictures by just using a mask and a text prompt. You can select specific areas of your image with the mask and then describe the changes you want using natural language. The app will then smartly edit the selected area of your image based on your instructions.

## Usage
- **Edit Image (required)**: Upload the image you want to edit, then mark the area you want to change.
- **Prompt (required)**: Type a description of the changes you want to make to the masked area.
- **Context Scale**: Higher values preserve more detail from the original image.

### Tips
- If the result isn't great, try marking a bigger area or lowering the *context scale*.
- Image will be scaled to max 1440(width/height), to avoid OOM and fix lowres/blurry result.
- You can check the mask image output by clicking at *Preprocessor data*.

## ComfyUI
As of this writing, ComfyUI integration isn't supported yet.  
You can follow updates here: <https://github.com/comfyanonymous/ComfyUI/pull/11304>

The author decided to retrain everything because there was a bug in the v2.0 release. 
Once that's done, ComfyUI support will soon be available.  
Please wait patiently while the author trains v2.1.

# Known Bugs
- Result is washed out and blurry outside masked area.
- Error when changing the guidance_scale to above 1.

## References
- **alibaba-pai - Controlnet Union v2.0**: <https://huggingface.co/alibaba-pai/Z-Image-Turbo-Fun-Controlnet-Union-2.0>
- **VideoX-Fun**: <https://github.com/aigc-apps/VideoX-Fun>
- **Tongyi-MAI**: <https://huggingface.co/Tongyi-MAI/Z-Image-Turbo>

## Version History
2025-12-14: Published to huggingface space.