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
Turn off cuda malloc by default when --fast autotune is turned on. (#10393)
5b80add comfyanonymous committed 1 year ago
Disable torch compiler for cast_bias_weight function (#10384)
9da397e comfyanonymous committed 1 year ago
execution: fold in dependency aware caching / Fix --cache-none with loops/lazy etc (#10368)
b1467da rattus128 committed 1 year ago
Do batch_slice in EasyCache's apply_cache_diff (#10376)
d8d60b5 Jedrzej Kosinski committed 1 year ago
Workaround for nvidia issue where VAE uses 3x more memory on torch 2.9 (#10373)
19b4661 comfyanonymous committed 1 year ago
fix(api-nodes): remove "veo2" model from Veo3 node (#10372)
bc0ad9b Alexander Piskun committed 1 year ago
refactor: Replace manual patches merging with merge_nested_dicts (#10360)
afa8a24 Faych committed 1 year ago
Fix order of inputs nested merge_nested_dicts (#10362)
493b81e Jedrzej Kosinski committed 1 year ago
api-nodes: fixed dynamic pricing format; import comfy_io directly (#10336)
7a88384 Alexander Piskun committed 1 year ago
Better memory estimation for the SD/Flux VAE on AMD. (#10334)
dfff7e5 comfyanonymous committed 1 year ago
Fix loading old stable diffusion ckpt files on newer numpy. (#10333)
e4ea393 comfyanonymous committed 1 year ago
Enable RDNA4 pytorch attention on ROCm 7.0 and up. (#10332)
c8674bc comfyanonymous committed 1 year ago