[Feature Request]: Multi-image prompt

[Ehplodor]

Is there an existing issue for this?

• [X] I have searched the existing issues and checked the recent builds/commits

What would your feature do ?

A new feature of MidJourney V4 (currently alpha) allows for mixing multiple images together (ReMix). Results look great.

It would be awesome to be able to mix multiple images together in a similar fashion through img2img.

Proposed workflow

1. Go to img2img
2. Load N images and/or include images URLs (local or web)
3. Optionally add some prompt / alternate prompts (allow no prompts at all too)
4. Generate

Additional information

MagicMix as implemented in @mpaepper 's implementation, does seem to allow, at least that has been reported in this comment, to re-infuse image latents at each step. This might allow in the end also to mix another/multiple image latents at a given time step ?

Also, relevant discussion : https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/4595

[leohu1]

Great idea, but I think this shoud be a extension.

[leohu1]

It's similar to https://github.com/Kahsolt/stable-diffusion-webui-prompt-travel Only that extension is prompt travel and what you want is image travel. Maybe it will be useful.