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

Add fast preview support for Wan models.
comfyanonymous committed 1 year ago
c37f15f
Don't try to use clip_fea on t2v model.
comfyanonymous committed 1 year ago
4bca736
Better wan memory estimation.
comfyanonymous committed 1 year ago
b6fefe6
More code reuse in wan.
comfyanonymous committed 1 year ago
fa62287
Slightly better wan i2v mask implementation.
comfyanonymous committed 1 year ago
0844998
WIP support for Wan I2V model.
comfyanonymous committed 1 year ago
4ced06b
Wan seems to work with fp16.
comfyanonymous committed 1 year ago
cb06e96
Fix missing frames in SaveWEBM node.
comfyanonymous committed 1 year ago
0c32f82
Update README.md (#6960)
Yoland Yan committed 1 year ago
189da37
Make wan work with all latent resolutions.
comfyanonymous committed 1 year ago
9a66bb9
ea0f939
Change wan rope implementation to the flux one.
comfyanonymous committed 1 year ago
f37551c
WIP support for Wan t2v model.
comfyanonymous committed 1 year ago
6302301
Cleanup some lumina te code.
comfyanonymous committed 1 year ago
f400760
ace899e
Remove some useless code.
comfyanonymous committed 1 year ago
aff1653
Bump ComfyUI version to v0.3.15
comfyanonymous committed 1 year ago
b50ab15
072db3b
Latest mac still has the black image bug.
comfyanonymous committed 1 year ago
a6deca6
Let all model memory be offloaded on nvidia.
comfyanonymous committed 1 year ago
41c30e9
f579a74
Add discord channel to support section. (#6900)
Robin Huang committed 1 year ago
d372725
Apparently directml supports fp16.
comfyanonymous committed 1 year ago
12da6ef
c5be423
0d4d922