instruct-pix2pix icon indicating copy to clipboard operation
instruct-pix2pix copied to clipboard

Published 20 hours ago • verified publisher icon timothybrooks

As Automatic1111 extension?

Open mykeehu opened this issue 2 years ago • 11 comments

I really like this solution and have been looking for something similar for a long time! Could you make this as an extension for Automatic1111? The gradio is a given there too. I would really like it!

mykeehu avatar Jan 24 '23 22:01 mykeehu

For now you can check out my GUI implementation:

https://www.reddit.com/r/StableDiffusion/comments/10kbyro/nmkd_stable_diffusion_gui_190_is_out_now/

I don't think A1111 supports Diffusers-based implementations at all at the moment, not sure though.

n00mkrad avatar Jan 25 '23 00:01 n00mkrad

https://github.com/Klace/stable-diffusion-webui-pix2pix

Here you go, this is a fork of the latest Auto1111 with crude support for it until someone does it better or I can polish it into the proper pipeline.

Klace avatar Jan 25 '23 01:01 Klace

Thanks for the links, I will check it out!

mykeehu avatar Jan 25 '23 06:01 mykeehu

Here is a (basic) extension implementation:

https://github.com/Klace/stable-diffusion-webui-instruct-pix2pix

Klace avatar Jan 25 '23 17:01 Klace

@Klace I saw today on Automatic the preparations to support webUI. Thank you very much and I look forward to hearing more, because this is another level where we don't have to deal with inpaint anymore, but can exchange things with words.

I'm already trying, I need to relearn prompting. I love it! And you can now install it from the WebUI, from the Extension tab! image

I've opened a thread here to see if more people will share their experiences.

mykeehu avatar Jan 25 '23 19:01 mykeehu

related - https://www.youtube.com/watch?v=ro1gJY1ysqQ&ab_channel=NerdyRodent

johndpope avatar Jan 27 '23 23:01 johndpope

@Klace I saw today on Automatic the preparations to support webUI. Thank you very much and I look forward to hearing more, because this is another level where we don't have to deal with inpaint anymore, but can exchange things with words.

I'm already trying, I need to relearn prompting. I love it! And you can now install it from the WebUI, from the Extension tab! image

I've opened a thread here to see if more people will share their experiences.

Im using a collab notebook and have put the extension into extensions folder and enabled insecure extensions but the instruct tab does not show up, any ideas why?

G-force78 avatar Jan 28 '23 11:01 G-force78

related - https://www.youtube.com/watch?v=ro1gJY1ysqQ&ab_channel=NerdyRodent

Great video, thanks!

mykeehu avatar Jan 28 '23 17:01 mykeehu

Just to follow up for anyone checking this thread I integrated support into the core webui via img2img, no need for an extension anymore. Using in img2img opens up almost all the existing auto1111 functionality, so don't use the extension anymore. :)

Klace avatar Feb 07 '23 17:02 Klace

Thank you for the fast and efficient solution, integration, I using it almost every day!

mykeehu avatar Feb 08 '23 19:02 mykeehu

@Klace Thank you for the integration. Am I using the right version of the webui/stable diffusion that has this support built-in? It doesn't seem to respond well to the prompt box at the top e.g. "convert her into a boy" Is there something else I need to do?

I am using this docker: https://hub.docker.com/r/holaflenain/stable-diffusion

I'm trying to get results like this:

image

I tried to match the CFG and steps value but instead, it's drawing something completely different:

image

frakman1 avatar Jul 02 '23 17:07 frakman1