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 Load Image Output node (#6790)
bymyself committed 1 year ago
afc85cd
acc152b
Fix typo.
comfyanonymous committed 1 year ago
b07258c
Improve AMD arch detection.
comfyanonymous committed 1 year ago
31e54b7
bf16 manual cast works on old AMD.
comfyanonymous committed 1 year ago
8c0bae5
Refactor torch version checks to be more future proof.
comfyanonymous committed 1 year ago
530412c
61c8c70
Update frontend to v1.9.18 (#6828)
Comfy Org PR Bot committed 1 year ago
d0399f4
Disable bf16 on AMD GPUs that don't support it.
comfyanonymous committed 1 year ago
e2919d3
93c8607
Update frontend to v1.9.17 (#6814)
Comfy Org PR Bot committed 1 year ago
b3d6ae1
2e21122
Disable pytorch attention in VAE for AMD.
comfyanonymous committed 1 year ago
1cd6cd6
Fix ruff.
comfyanonymous committed 1 year ago
1d5d658
mix_ascend_bf16_infer_err (#6794)
zhoufan2956 committed 1 year ago
3574025
Add add_weight_wrapper function to model patcher.
comfyanonymous committed 1 year ago
ab888e1
Cleanup.
comfyanonymous committed 1 year ago
d9f0fcd
Fix for running via DirectML (#6542)
HishamC committed 1 year ago
b124256
e57d228
Make lumina model work with any latent resolution.
comfyanonymous committed 1 year ago
4027466
Remove useless function.
comfyanonymous committed 1 year ago
095d867
Make error more clear to user.
comfyanonymous committed 1 year ago
3d06e1c
Allow FP16 accumulation with `--fast` (#6453)
catboxanon committed 1 year ago
43a74c0
Document which text encoder to use for lumina 2.
comfyanonymous committed 1 year ago
af93c8d