I'm unsure if ComfyUI requirements preclude this, but I find it a little annoying that the template doesn't allow me to cache the base SD models in my network storage for faster boot. Secondarily, I don't plan to use the base models and have populated my rsync'd models directory with the models I intend to use
- Please make the base model download for the ComfyUI template configurable/optional
- Please move the download of the required base models out of the Dockerfile and have the entrypoint script check the network storage for their presence before forcing a 10+gb download each time a pod is started with this template