Update diffusers to support InstructPix2Pix pipeline

[Hades32]

https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/pix2pix

[gadicc]

Thanks for the PR, @Hades32!

I've also had my eye on this one for a while `:) So thanks for this.

I'll be merging a lot of stuff into dev later this week (everything from the split branch, to finally move forward with "v1"), and will rebase this into that straight afterwards. I'll keep you posted!

Thanks again :)

[gadicc]

Ok I moved up my timetable a bit :sweat_smile: Merged everything to dev :muscle:

The resulting merge conflict from your PR wasn't terrible but I wanted to update to the latest diffusers commit hash too, so unfortunately I won't get to merge this... sorry about that... but you really did succeed in nudging me to move things forward :raised_hands:

Just note that everything I merged today bumps the dev branch from v0 to v1 (not officially released yet, but getting there). There is a dev-v0-final branch pointing to the last dev commit of v0 if relevant. Otherwise see Upgrading from v0 to v1.

For latest diffusers (in my commit https://github.com/kiri-art/docker-diffusers-api/commit/87632aa2c32faddfeb049fe969884b568066edd3), you can either build from source in dev or simply use the docker image with dev tag, i.e. docker pull gadicc/diffusers-api:dev. On next merge to main, it will get tagged with appropriate semantic version too.

Hope that's all clear, not sure if you're ready to try out v1 yet since it's not officially released... but I'm available for anything you need. Thanks for your continued support! :pray:

[Hades32]

Haha, it was just a single line, but very happy to help give some extra motivation :D

Thanks - as always - for your great work! I'll let you know when I stumble over any issues

[gadicc]

🙏