AI Clipping ComfyUI
https://github.com/anil-matcha/ai-clipping-comfyui- Category
- AI Tools
- Rank
- No. 1539Tools index
Previous survey · No. 1553 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- Anil-matcha
- GitHub
- 22 stars
- Date
About
Free open-source Opus Clip alternative for ComfyUI — turns long-form videos and podcasts into ranked viral shorts with face-tracked auto-crop.
What it does
A ComfyUI custom-node pack that sends a hosted video or uploaded local file to a managed clipping service. It polls for completion, exposes selected moments and editorial metadata, previews the leading result, and can download clips as frame tensors for downstream graph processing.
Why it's ranked here
This is a focused integration with useful workflow boundaries. It handles submission, polling, result selection, previewing, and saving inside ComfyUI, while coordinates-only output lets users keep rendering in their own pipeline. Its value depends heavily on accepting a remote paid service.
What's good
The nodes compose cleanly. Users can fan results into separate branches, inspect titles and hook text, save videos, or pass frames onward. Three aspect ratios cover common portrait and feed formats. Local files upload automatically, while hosted URLs avoid that transfer step.
Tradeoffs
The repository is open source, but transcription, selection, deduplication, and cropping happen on muapi.ai. An API key and sufficient credits are therefore required. The main request may poll for up to thirty minutes. Frame previews and saved-frame tensors also require downloading and decoding video locally.
How to use it well
Best for ComfyUI users building repeatable short-form production graphs around long recordings. Use coordinates-only output when another renderer should control the final cut. Use the saved frames for later visual processing. Look elsewhere for a broad model collection or a standalone application without ComfyUI.
Technical notes+
__init__.py merges node mappings from ai_clipping_nodes.py and ai_clipping_video_saver.py. The main node posts to /api/v1/ai-clipping, polls /api/v1/predictions/{request_id}/result every 10 seconds, and stops after 1,800 seconds. It accepts several backend response shapes before normalizing clips. Local sources upload through /api/v1/upload_file. ai_clipping_video_saver.py streams the returned MP4 into ComfyUI's output directory, finds a nonconflicting numbered name, decodes frames with OpenCV, and returns a Torch tensor. requirements.txt declares requests, Pillow, NumPy, Torch, and OpenCV. AiClipping_Example.json and AiClipping_FanOut_Example.json provide single-result and branched workflows.
Observed
- License
- MIT License
- Primary language
- Python
- Interface
- ComfyUI custom nodes backed by the muapi.ai HTTP API
- Installation
- ComfyUI Manager via Git URL, or manual clone plus pip requirements installation
- Runtime requirement
- Python 3.8 or newer
- Input support
- Hosted video URL or local video file upload
- Output formats
- Rendered clips or coordinates-only highlight time ranges
- Declared dependencies
- requests, Pillow, NumPy, Torch, and OpenCV
Read from README.md, requirements.txt, __init__.py, ai_clipping_nodes.py, ai_clipping_video_saver.py, LICENSE, AiClipping_Example.json, AiClipping_FanOut_Example.json.
What it can do
Convert long-form videos into short clips
Long-form video file → Multiple short video clips
Convert podcasts into short clips
Podcast audio/video file → Multiple short video clips
Rank generated clips by viral potential
Generated short clips → Ranked list of clips ordered by viral potential
Auto-crop videos with face tracking
Video with faces → Cropped video focused on tracked faces
Extract highlight segments from long content
Long-form video or podcast → Key highlight segments
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.