dream-textures
dream-textures copied to clipboard
AUTOMATIC1111 version works on 2GB
Hi Carson,
Actually, I was able to make SD work on my machine from here https://github.com/AUTOMATIC1111/stable-diffusion-webui/
With these arguments in webui-user.bat
set COMMANDLINE_ARGS=--lowvram --precision full --no-half
So I guess you can make your version work as well by implanting this in the settings; thus more people can enjoy your add-on (mostly laptops)
I'd like to mention that their lowvram setting appears to use the CPU instead of the GPU.
https://github.com/AUTOMATIC1111/stable-diffusion-webui/blob/405c8171d1acbb994084d98770bbcb97d01d9406/modules/processing.py#L391-L392
Oh man! Don't tease me... I have an old 2gb CUDA card and I would love to be able to use it for this...PLEASE!?!?!?
@rexorp based on what NullSense linked, it will not run on 2GB and simply switches to your CPU when you don’t have enough.
Collaborator
Thank you for the clarification, I thought it will still use GPU but it will be just a bit slower... I hope that someday it will work with low VRAM since, I've seen some Researchers talking about optimizing things to use much less...
Hi. I have a GPU with 4GB vRAM. I'm using "lowvram" argument and it definitely uses my GPU (Monitored with nvidia-smi on Linux). I am able to generate images bigger than 512x512 with lowvram argument, which without it I can't go beyond 384x384 at most.
Edit: Here the explanation, it uses CPU and GPU: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Optimizations
Anyway, a similar implementation in Dream Textures would be nice for those with 4GB vRAM or less.
OMG it's happening, everybody calm down!
One API for krita, blender, inkscape, ... , would be nice.
https://github.com/AUTOMATIC1111/stable-diffusion-webui/pull/765
Is this really a BUG of DT add-on for Blender?
This issue is stale because it has been open for 60 days with no activity.
Guys try Docker Version it uses GPU and works pretty well as is (I didn't test with Dream-Textures yet)
docker-compose.override.yml
services:
auto:
environment:
- CLI_ARGS=--allow-code --lowvram --no-half --xformers --opt-split-attention --enable-insecure-extension-access --precision full --api
I will find time to do the testing... it is hard with all the crazy stuff going on lately... and comeback with some more tips and results.
This issue is stale because it has been open for 60 days with no activity.
This issue was closed because it has been inactive for 7 days since being marked as stale.