Skip to content

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

Reduce tiled decode peak memory (#13050)
Jukka Seppänen committed 8 months ago
fd0261d
f6b869d
Fix regression. (#13053)
comfyanonymous committed 8 months ago
56ff88f
Further Reduce LTX VAE decode peak RAM usage (#13052)
Jukka Seppänen committed 8 months ago
9fff091
b67ed2a
cad24ce
68d542c
bump manager version to 4.1b6 (#13022)
Dr.Lt.Data committed 8 months ago
8b9d039
chore: update workflow templates to v0.9.26 (#13012)
ComfyUI Wiki committed 8 months ago
379fbd1
fix: disable SageAttention for Hunyuan3D v2.1 DiT (#12772)
Paulo Muggler Moreira committed 8 months ago
8cc746a
Fix potential issue. (#13009)
comfyanonymous committed 8 months ago
ca17fc8
7a16e8a
Skip running model finalizers at exit (#12994)
blepping committed 8 months ago
b202f84
feat(assets): align local API with cloud spec (#12863)
Luke Mino-Altherr committed 8 months ago
2bd4d82
Enable Pytorch Attention for gfx1150 (#12973)
lostdisc committed 8 months ago
3814bf4
Make EmptyLatentImage follow intermediate dtype. (#12974)
comfyanonymous committed 8 months ago
d062bec