comfyui
https://github.com/comfyanonymous/comfyui
Python
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported24 Subscribers
Add a CodeTriage badge to comfyui
Help out
- Issues
- Qwen-Image 2.1: --enable-dynamic-vram silently corrupts output after any model reload (channel slice offset, ROCm gfx1201)
- Qwen-Image-2.1 image edit: VAE reference-latent splice produces broadband noise at exactly resolution=1024 (the node default); 992 and 1056 are clean, reproduces on CPU too
- Support NVFP4 conditioning for Qwen Image 2.1
- Fix default inpainting workflow to preserve unmasked pixels
- README: tell NVIDIA users what to do about "no kernel image is available"
- Fix Qwen Image PEFT LoRA mapping
- YuE2 For some reason, only one job runs.
- Quantized (W4A8) H3 + LoRA patches adaln_proj while weight is still in packed shape (v0.36.0)
- Use cuda_device_context when partially loading models
- Add an opt-out for the auto-enabled fast-disk policy (7a0b5ee): on high-RAM machines it streams weights from NVMe on every step
- Docs
- Python not yet supported