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
LTXV: Accumulate VAE decode results on intermediate_device (#12955)
0904cc3 Jukka Seppänen committed 8 months ago
Update comfyui-frontend-package to version 1.41.20 (#12954)
4941cd0 comfyanonymous committed 8 months ago
Add --fp16-intermediates to use fp16 for intermediate values between nodes (#12953)
c711b8f comfyanonymous committed 8 months ago
fix: use no-store cache headers to prevent stale frontend chunks (#12911)
e0982a7 Christian Byrne committed 8 months ago
comfy aimdo 0.2.11 + Improved RAM Pressure release strategies - Windows speedups (#12925)
7810f49 rattus committed 8 months ago
fix(api-nodes): Tencent TextToModel and ImageToModel nodes (#12680)
f9ceed9 Alexander Piskun committed 8 months ago
Revert "Revert "feat: Add CacheProvider API for external distributed caching"" (#12915)
4a8cf35 Deep Mehta committed 8 months ago
Revert "feat: Add CacheProvider API for external distributed caching (#12056)" (#12912)
d1d53c1 comfyanonymous committed 8 months ago
feat: Add CacheProvider API for external distributed caching (#12056)
af7b4a9 Deep Mehta committed 8 months ago
fix: use frontend-compatible format for Float gradient_stops (#12789)
3fa8c56 Terry Jia committed 8 months ago
Support flux 2 klein kv cache model: Use the FluxKVCache node. (#12905)
44f1246 comfyanonymous committed 8 months ago
Fix issue with batch_size > 1 on some models. (#12892)
f6274c0 comfyanonymous committed 8 months ago
fix: guard torch.AcceleratorError for compatibility with torch < 2.8.0 (#12874)
4f4f865 Adi Borochov committed 8 months ago