Halo-Lipsy is a native Wav2Lip implementation for ComfyUI that finally works
on AMD APUs with unified memory (Strix Halo, etc.).
Key features:
- No subprocess hacks - runs natively in ComfyUI process
- Face detection on CPU to prevent memory conflicts
- Safe tensor casting for unified memory compatibility
- Sync tuning and edge blending options
- Works on ROCm, AMD APUs, and NVIDIA
Tested on Ryzen AI Max+ 395 (Strix Halo) with 64GB unified memory and ROCm 7.11.
Repository: https://github.com/bkpaine1/Halo-Lipsy
Co-authored-by: bkpaine1 <bkpaine1@users.noreply.github.com>
* Add ComfyUI_FMJ_SaveImageVersions node entry
Save images with full metadata (prompt, workflow, software versions, commit)
This repository provides ComfyUI nodes to:
- Save PNG images with embedded metadata (positive / negative prompts and other JSON fields),
- Generate and copy a snapshot describing the environment (commits, versions, GPU, etc.),
- Load an image and restore information from its metadata.
Important: this project runs a local script `snapshot.py` to capture the state of ComfyUI and custom nodes. Only run it in trusted environments.
* Update custom-node-list.json
---------
Co-authored-by: Dr.Lt.Data <128333288+ltdrdata@users.noreply.github.com>
* Update ComfyUI-LLM-Nodes to ComfyUI-Gemini-LiteLLM v3.0.0
- Rename repository from ComfyUI-LLM-Nodes to ComfyUI-Gemini-LiteLLM
- Remove OpenAI support, Gemini-only implementation
- Support multimodal chat and image generation
- Add temperature control (0-1) and aspect ratio options
- Update node ID and description in custom-node-list.json
- Update repository URL in extension-node-map.json
- Remove OpenAIImageParams from node list
* Fix description text in custom-node-list.json
* Fix formatting in custom-node-list.json
Dr.Lt.Data