"Prompt-to-Prompt Image Editing with Cross Attention Control"

[lendrick]

Saw this in a thread posted on reddit the other day:

https://www.reddit.com/r/StableDiffusion/comments/x98py5/reproducing_the_method_in_prompttoprompt_image/

I recommend checking out the paper. Apparently some really amazing stuff is possible with this.

https://arxiv.org/abs/2208.01626

There's someone working on implementing it. I'm not sure how far they've gotten or if they have a github repo yet, but this is definitely something worth keeping an eye on.

[831Digital]

It has an implementation here: https://github.com/bloc97/CrossAttentionControl

Would love to see this added.

[mchaker]

(bumping, upvoting, thumbs up with both hands and feet)

[HalfMAI]

Try using this commit
Doggettx/stable-diffusion@ccb17b5

Result:

[HalfMAI]

Made a pull request(#1229)
But I'm very new to this repo, hope someone more familiar can help