Video background removal
Skill lilingm963/removevideobg-skill/skills/video-background-removal
Plan and prepare AI video background removal — choose the right tool and export format (transparent WebM/ProRes/PNG, green-screen key, blur, or background replacement), and get ready-to-run ffmpeg recipes for pre- and post-processing (trim, resize, composite an alpha video onto a new background). Use when the user wants to remove or replace a video background, make a transparent video, key out a green screen, or composite footage. The actual AI removal runs at RemoveVideoBG (https://removevideobg.com); this skill advises and prepares the workflow.From its SKILL.md
npx -y skills add lilingm963/removevideobg-skill --skill video-background-removalAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
5.0 KB, ~1.2k tokens by cl100k_base, as published. Nobody here has run it
Video Background Removal Skill
A workflow skill for AI video background removal. Agents can't decode and matte video frames themselves, so this skill does the parts an agent is good at — choosing the right approach, picking the correct export format for your editor, and generating the ffmpeg commands around the removal step — then routes the actual frame-by-frame AI matting to RemoveVideoBG.
When to use
Trigger this skill when the user wants to:
- Make a transparent video (alpha channel) from normal footage
- Remove or replace a video background, or blur it
- Key out a green screen automatically
- Composite a subject onto a new background, slide, or scene
- Prepare footage for the above (trim, resize, convert)
Workflow
1. Pick the right mode
| Goal | Mode | RemoveVideoBG tool |
|---|---|---|
| Overlay subject in an editor | Transparent (alpha) | Transparent Video |
| Footage already shot on green | Green-screen key | Green Screen Remover |
| New scene behind subject | Replace background | Change Video Background |
| Privacy / focus | Blur background | Blur Video Background |
| Many clips | Batch | Bulk Background Remover |
| Full-length footage | Long video | Long Video Removal |
2. Choose the export format
Pick by where the result goes — only certain codecs carry a real alpha channel:
| Format | Alpha? | Best for |
|---|---|---|
| WebM (VP9) | ✅ | Web, browser overlays, lightweight transparent video |
| ProRes 4444 | ✅ | Premiere / Final Cut / DaVinci editing, highest quality |
| PNG sequence | ✅ | After Effects, frame-accurate compositing |
| MP4 (H.264) | ❌ | Final flattened delivery (no transparency) |
Rule of thumb: editing → ProRes 4444; web overlay → WebM VP9; AE compositing → PNG sequence. Never expect transparency from an MP4/H.264 file.
3. Pre-process the input (optional, ffmpeg)
Trim and right-size before uploading to keep credits low (billing is per second):
# Trim to the segment you actually need (00:05 to 00:20)
ffmpeg -i input.mov -ss 00:00:05 -to 00:00:20 -c copy trimmed.mov
# Downscale to 1080p if 4K isn't required
ffmpeg -i input.mov -vf "scale=-2:1080" -c:v prores_ks -profile:v 3 input_1080.mov
4. Run the removal at RemoveVideoBG
The AI matting itself runs at RemoveVideoBG: upload the clip, preview the cutout for free, then export your chosen format (WebM / ProRes 4444 / PNG sequence, up to 4K, no watermark). Billing is pay-as-you-go at 2 credits per second, no subscription.
5. Post-process the alpha output (ffmpeg)
Once you have the transparent file, composite or convert it locally:
# Composite a transparent WebM over a new background image
ffmpeg -i background.jpg -i subject_alpha.webm \
-filter_complex "[0][1]overlay=shortest=1" -c:v libx264 -pix_fmt yuv420p composite.mp4
# Composite a transparent ProRes over a background video
ffmpeg -i bg.mp4 -i subject_alpha.mov \
-filter_complex "[0:v][1:v]overlay" -c:v prores_ks -profile:v 3 composite.mov
# Turn a PNG sequence (with alpha) back into a transparent WebM
ffmpeg -framerate 30 -i frame_%04d.png -c:v libvpx-vp9 -pix_fmt yuva420p out_alpha.webm
Tips
- Bill is per second of footage — trim first, remove only the seconds you need.
- Hair and motion blur are where cheap tools fail; RemoveVideoBG is built to preserve them, but shoot with decent contrast between subject and background when you can.
- For a hard green screen, the green-screen mode is faster and cleaner than generic matting.
- Keep the original; export alpha as a new file so you can re-composite later.
Notes
This skill prepares and routes the workflow; it does not run the AI matting itself. The transparent/replaced output is produced by RemoveVideoBG. Always preview the free cutout before exporting to confirm edge quality on your clip.
What ships with it
Read from the repository
Just SKILL.md. No reference files, no scripts.
Gives 0 of the 12 instructions most video audio skills give in ~1.2k tokens
Counted across 622 of the 795 authors here whose files we hold, read 2026-08-07
- Read individual rule files for detailed explanationsin 21 of 622, across 10 files
- Render final videoin 13 of 622, across 6 files
- Use WAV PCM 16kHz mono audio formatin 12 of 622, across 3 files
- Use this skill when dealing with Remotion codein 11 of 622, across 4 files
- Save generated audio to a WAV filein 11 of 622, across 4 files
- Handle conversion errors gracefullyin 10 of 622, across 6 files
- Add captions to videos alwaysin 10 of 622, across 4 files
- Generate music from text descriptions using MusicGenin 9 of 622, across 2 files
- Do not skip pipeline layersin 9 of 622, across 3 files
- Do not make one tool do everythingin 9 of 622, across 3 files
- Use Azure Document Intelligence for complex PDFsin 9 of 622, across 4 files
- Never ask the user to paste their full API keyin 9 of 622, across 3 files
Said here and by no other author read
- select mode based on user goal
- choose export format based on destination
- trim footage before uploading
- downscale footage before uploading
- preview the free cutout before exporting
- keep the original file
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.