stable-diffusion-rocm VS stable-diffusion

Re: stable diffusion https://github.com/AshleyYakeley/stable-diffusion-rocm

Very easy actually. This is not officially documented, but with a recent enough kernel you don't have to install anything. You can grab the official rocm container and it'll just work. For example for Stable Diffusion see https://github.com/AshleyYakeley/stable-diffusion-rocm/blob/...

I had good luck with these directions, which let you run inside a docker container:

https://github.com/AshleyYakeley/stable-diffusion-rocm

I had to make the one line change suggested in issue #3 to get it to run under 8GB.

radeontop suggests 4GB might work.

I also had to add this environment variable to make it work on my unsupported radeon 6600xt:

HSA_OVERRIDE_GFX_VERSION=10.3.0

It takes under two minutes per batch of 5 images with the --turbo option.

(Base OS is manjaro; using the distro's version of docker; not the flatpack docker package.)

If you don't have a GPU, paperspace will rent you an appropriate VM.

I have it working on an RX 6800, used the scripts from this repo[0] to build a docker image that has ROCm drivers and PyTorch installed.

I'm running Ubuntu 22.04 LTS as the host OS, didn't have to touch anything beyond the basic Docker install. Next step is build a new Dockerfile that adds in the Stable Diffusion WebUI.[1]

[0] https://github.com/AshleyYakeley/stable-diffusion-rocm

use or update this one : https://github.com/hlky/stable-diffusion it has all the samplers, and if you want perfect faces, try k_euler_a

try this one https://github.com/hlky/stable-diffusion you need at least a 1050 to run it tho

https://github.com/hlky/stable-diffusion/issues/153 With 36 comments and tons of before and after comparisons, which are now deleted

I wanted to try hlky version (https://github.com/hlky/stable-diffusion) , due to the WebUI and integration with upscaling models. It should also have the option to be optimized for low VRAM. To avoid getting a green square I have to add the parameters "--precision full --no-half". When I run a prompt, even with the smallest image size, I immediately get a CUDA memory error. Interestingly, without these parameters there isn't any memory error (but, of course, the result is a green square)

Made using https://github.com/hlky/stable-diffusion

Any chance to get this integrated into something like hlky's web ui?

Linux Guide her. Jeg har også Linux, men jeg valgte å sette det opp på Windows boksen min fordi driverne til Nvidia kortet på Linux ikke er helt sammarbeidsvillig når det kommer til å justere viftene etter sensorene i kortet (så jeg må sette det manuelt).

I'm seeing GFPGAN essentially remove all texture from faces, and I only want to use it on the eyes. Any thoughts on how to do this? I am using hlky/stable-diffusion now but I have no issues running a different repo/fork if needed and using command line.