Video Background Remover CLI
github.com
- Free plan
- Windows
- Linux
- Self-hosted
What Video Background Remover CLI is
Video Background Remover CLI is a free Python command-line tool for removing video backgrounds with rembg and OpenCV. It accepts MP4 and MOV input and can export processed video, transparent PNG or WebP frames, and animated WebP or GIF files. Removed backgrounds can be replaced with a solid color or an image. Available models include isnet-general-use, u2net, u2netp, u2net_human_seg, and silueta. The CLI can convert MatAnyone foreground and alpha video pairs or invoke MatAnyone 2 through matanyone2-runtime. A bundled Gradio WebUI includes a CLI Export tab and helpers for WebP, GIF, frame ZIPs, MP4, and WebM. Installation is documented with pip install video-background-remover. Python 3.10 or later is required, FFmpeg is not required, and the model downloads on first run. Regular video output composites transparent pixels onto a selected background or black rather than preserving alpha. The README notes that long animated GIFs can become large, while silueta favors quality over speed and lightweight u2netp may lose quality on complex fire frames.
Who it suits
This tool suits Python users who want command-line or Gradio workflows for removing backgrounds from MP4 or MOV clips. It also fits users needing frame, GIF, WebP, or replacement-background outputs.
Good to know, and what to watch
- Free Python tool with pip installation
- Supports MP4 and MOV input
- Offers several removal models
- Includes a Gradio WebUI
- Does not require FFmpeg
- Requires Python 3.10 or later
- Regular video output does not preserve alpha
- Long animated GIFs can become large
- Model quality and speed vary
Plans
Every plan and priceVideo Background Remover CLI
The verdict
Video Background Remover CLI offers model-based background removal with both command-line and web interface options. Its output formats are broad, but transparent video requires compositing and model selection involves quality and speed tradeoffs.
What the maker publishes
- Animated exports
- It can create animated WebP, GIF, or both from video. source
- API availability
- The published documentation exposes command-line and Python module entry points but no hosted HTTP API. source
- Background options
- Users can replace the removed background with a solid color or an image. source
- Background replacement
- Removed backgrounds can be replaced with a solid color or an image. source
- CLI entry points
- The package exposes video-background-remover, vbr, video-background-remover-webui, and vbr-webui commands. source
- Default model
- isnet-general-use is the documented default model. source
- Developer dependencies
- The project declares rembg, OpenCV, NumPy, Pillow, tqdm, onnxruntime, and imageio-ffmpeg dependencies. source
- FFmpeg requirement
- The documentation says FFmpeg is not required. source
- First-run model download
- Models download automatically on first use and are cached locally. source
- GIF limit
- The README cautions that animated GIF exports of long videos can become large. source
- Green-fringe cleanup
- MatAnyone processing includes cleanup for green matte contamination around semi-transparent edges. source
- Installation
- The package can be installed with pip as video-background-remover. source
- Install method
- The README documents installation from PyPI with pip install video-background-remover. source
- Licence status
- The repository file listing and pyproject metadata show no declared software licence. source
- Limit: GIF size
- The README warns that long animated GIF exports can become large. source
- Limit: processing startup
- The documentation warns that initial model loading can take time. source
- Limit: regular video alpha
- Regular MP4 output does not retain alpha transparency and composites onto a selected background or black. source
- Maker identity
- The GitHub organization profile identifies the account as Maki Sunwood-ai-labs and links to hamaruki.com. source
- Maker portfolio
- Devpost lists Maki Maki (sunwood-ai-labs) with one project and one hackathon. source
- Maker website date
- The maker's website footer states © 2022 Sun wood AI labs.2. source
- MatAnyone backend
- It can invoke matanyone2-runtime through a separately specified Python interpreter. source
- MatAnyone integration
- The CLI can convert MatAnyone foreground and alpha video pairs, or invoke MatAnyone 2 through matanyone2-runtime. source
- MatAnyone pairs
- The tool combines matching foreground and alpha video files into transparent outputs. source
- Model selection
- Available rembg models are isnet-general-use, u2net, u2netp, u2net_human_seg, and silueta. source
Checked against the maker's pages on 21 Oct 2026.