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

Refactor the attention stuff in the VAE.
comfyanonymous committed 3 years ago
23680a9
c8013f7
Fix node getBounding for collapsed nodes
Jairo Correa committed 3 years ago
5a608aa
682c84c
Fix add selected nodes to empty group
Jairo Correa committed 3 years ago
e8c0221
Group options
Jairo Correa committed 3 years ago
7d5d0fd
Add a separate optimized_attention_masked function.
comfyanonymous committed 3 years ago
bb064c9
Make clear that the old CheckpointLoader is deprecated.
comfyanonymous committed 3 years ago
7e09e88
Group menu option select nodes
Jairo Correa committed 3 years ago
a7b65b9
Allow all extensions if extension list is empty
Jairo Correa committed 3 years ago
8d04978
Stop auto queue on error
Jairo Correa committed 3 years ago
2e6270e
25f0f4e
Add a --bf16-unet to test running the unet in bf16.
comfyanonymous committed 3 years ago
fd4c5f0
9a55dad
Fix MaskComposite
Jairo Correa committed 3 years ago
b5fa3d2
Fix FeatherMask
Jairo Correa committed 3 years ago
87097a1
Move and comment out.
comfyanonymous committed 3 years ago
fee3b0c
add query
Chris committed 3 years ago
41d2c56
Pull some small changes from the other repo.
comfyanonymous committed 3 years ago
20d3852
Allow attn_mask in attention_pytorch.
comfyanonymous committed 3 years ago
ac7d8cf
Refactor the attention functions.
comfyanonymous committed 3 years ago
1a4bd9e