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

Bump ComfyUI version to v0.3.12
comfyanonymous committed 1 year ago
55add50
Fix some cosmos fp8 issues.
comfyanonymous committed 1 year ago
0aa2368
cca96a8
Bump ComfyUI version to 0.3.11
comfyanonymous committed 1 year ago
619b8cd
Code refactor.
comfyanonymous committed 1 year ago
31831e6
Tweak hunyuan memory usage factor.
comfyanonymous committed 1 year ago
88ceb28
Clean up some debug lines.
comfyanonymous committed 1 year ago
23289a6
9d8b6c1
Slightly lower hunyuan video memory usage.
comfyanonymous committed 1 year ago
6320d05
Lower cosmos diffusion model memory usage.
comfyanonymous committed 1 year ago
25683b5
Lower cosmos VAE memory usage by a bit.
comfyanonymous committed 1 year ago
4758fb6
Optimize first attention block in cosmos VAE.
comfyanonymous committed 1 year ago
0087611
3.13 doesn't work yet.
comfyanonymous committed 1 year ago
bfd5dfd
Remove python 3.8 from test-build workflow.
comfyanonymous committed 1 year ago
55ade36
Add minimum numpy version to requirements.txt
comfyanonymous committed 1 year ago
2e20e39
3baf92d
Use latest python 3.12.8 the portable release.
comfyanonymous committed 1 year ago
1709a84
Remove unsafe embedding load for very old pytorch.
comfyanonymous committed 1 year ago
cba58ff
2feb8d0
5b657f8
Add SetFirstSigma node (#6459)
catboxanon committed 1 year ago
2cdbaf5
3aaabb1
Remove useless code.
comfyanonymous committed 1 year ago
1f1c7b7
Add res_multistep sampler from the cosmos code.
comfyanonymous committed 1 year ago
90f349f
Merge ruff.toml into pyproject.toml (#6431)
Chenlei Hu committed 1 year ago
42086af
Fast latent preview for Cosmos.
comfyanonymous committed 1 year ago
ee8a7ab
Add pyproject.toml (#6386)
Chenlei Hu committed 1 year ago
9c773a2