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
Lower memory usage for loras in lowvram mode at the cost of perf.
db8b59e comfyanonymous committed 2 years ago
Replace prints with logging and add --verbose argument.
65397ce comfyanonymous committed 2 years ago
Move some stable cascade nodes outside of _for_testing.
55f37ba comfyanonymous committed 2 years ago
Set upscale algorithm to bilinear for stable cascade controlnet.
03e6e81 comfyanonymous committed 2 years ago
Add debugging info for when comfy_extra nodes fail to import.
a38b9b3 comfyanonymous committed 2 years ago