Assuming SD models are converted to the onnx format (or whatever its called) shouldn't current sd projects offer similar performance? The dominant cost is in generation is inference, not the project runtime I thought.
I think the purpose isn't to improve performance but to make it not depend on a bunch of Python packages, which can be difficult to install.
This. Conda and Venv in paper should be like docker plug and play... But in practice they can be a hit or miss. Still better than manual global library management.
Conda requires a commercial license too
And is a 900MB download on Widndows.
Apparently binaries take space, what a surprise. /s
Fwiw it's not impossible to package all this junk up in Docker. I did it recently. You have to run Docker w/ a whole bunch of flags though, like `--gpus all --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 -p 7860:7860`. And the GPU isn't available during build time unless you do some hacks. But you can build the CUDA junk into the docker image along with Python and all the libs. Combine that with a .bat or .sh to contain the giant `docker run` command and you have a half-OK solution. Aside from having to start docker before being able to run it.
That's a nice tip. I'll try it. I had faced issues trying to run Automatic 1111 + oogabooga webui because of conda conflicts.