The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Commits on master
Merge branch 'patch-1' of https://github.com/savolla/ComfyUI
257c2ea comfyanonymous committed 2 years ago
--disable-smart-memory now unloads everything like it did originally.
a252963 comfyanonymous committed 2 years ago
Greatly improve lowvram sampling speed by getting rid of accelerate.
36a7953 comfyanonymous committed 2 years ago
Fix potential tensor device issue with ImageCompositeMasked.
6781b18 comfyanonymous committed 2 years ago
LoadImage now loads all the frames from animated images as a batch.
a1e1c69 comfyanonymous committed 2 years ago
Merge branch 'test-reliability' of https://github.com/pythongosssss/ComfyUI
ba3f3aa comfyanonymous committed 2 years ago
Fix SDXL area composition sometimes not using the right pooled output.
8cf1daa comfyanonymous committed 2 years ago
Add --deterministic option to make pytorch use deterministic algorithms.
2f9d6a9 comfyanonymous committed 2 years ago
Fix name counter preventing more than 3 of the same node Fix linked widget offset when populating values
6453dc1 pythongosssss committed 2 years ago
Don't resize clip vision image when the size is already good.
e45d920 comfyanonymous committed 2 years ago
Set clip vision model in eval mode so it works without inference mode.
719fa08 comfyanonymous committed 2 years ago