Local image studio · MOSS creates images, Qwen Turbo edits an uploaded image.
↥
Add a reference image
Choose a photo from your device or drag one here.
↥
Add a second image (optional)
Qwen can combine this with image 1 — describe how in your prompt.
↥
Add a third image (optional)
Qwen can combine this with images 1 and 2 — describe how in your prompt.
Your source images are kept unchanged. Image 1 is required, images 2 and 3 are optional. In your prompt, say how to combine them (e.g. “put the character from image 1 into the scene from image 2”).
Before / after
9:16 · Portrait
768 × 1344 px · Approx. 1.0 MP
This is the shape of the MOSS output canvas, not a generated image.
INPUT
Upload an image in Qwen mode
LATEST RESULT
Your result appears here
For MOSS, the input side is optional. For Qwen, compare your source and latest result.
Advanced — steps, CFG, size, seed
Ratio + size set the width/height above (rounded to 64). You can still edit width/height by hand — Generate uses those numbers. Qwen Edit ignores size, it follows your uploaded photo.
Multiple images are queued in ComfyUI and run one after another on the GPU — the GPU runs queued jobs one at a time, but you don't have to wait and press Generate again.
Past generations on this PC — including ones made in ComfyUI directly, not just here — newest first, as tiles. Tap a tile to view it full-size, or load it as a Qwen input. Delete all clears generated outputs only, not your uploaded photos.