![]() ![]() You can now launch ComfyUI with the command: python main.py Pip install torch torchvision torchaudio -extra-index-url xformers -r requirements.txt Then run the follow commands one by one: git clone Windows, open your Command Prompt (Search for "cmd"). Clone from Github (Windows, Linux) NVIDIA GPU ![]() You will get a folder called ComfyUI_windows_portable containing the ComfyUI folder.ĭouble click the file run_nvidia_gpu.bat to run with NVIDIA GPU, or run_cpu.bat to run with CPU. ![]() CPU generation is very slow, so this may as well be a NVIDIA-only installer:Ĭlick this link and your download will start: There is a portable standalone version of ComfyUI that will let you run ComfyUI with NVIDIA GPU or CPU. By learning how to use ComfyUI, you'll learn how Stable Diffusion actually works! How to install ComfyUI Quick-Install (Windows, NVIDIA) Learn how Stable Diffusion worksĬomfyUI is set up so that nodes reflect how a different part of the diffusion process works underneath the hood. This greatly optimizes the speed, with people reporting 3-5x faster generations with ComfyUI compared to AUTOMATIC1111. Why would you use such a complicated tool to do the same thing?īecause they are so configurable, ComfyUI generations can be optimized in ways that AUTOMATIC1111 generations cannot. ComfyUI VS AUTOMATIC1111Ī lot of newcomers to ComfyUI are coming from much simpler interfaces like AUTOMATIC1111, InvokeAI, or SD.Next. By creating and connecting nodes that perform different parts of the process, you can run Stable Diffusion. In ComfyUI, every node represents a different part of the Stable Diffusion process.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |