CHROMA KEY + AI PERSON MASK · VIDEO STUDIO · WAVE 24
Video Background Remover & Chroma Key Studio
Choose precise green/blue-screen removal or AI portrait segmentation for videos featuring people against ordinary backgrounds. Replace backdrops, refine masks, preview live and export supported WebM video.
Remove a green or blue background from video
Import a video shot against an evenly lit chroma-key screen. The processing engine compares each pixel with your chosen color, applies a tolerance and feathering threshold, composites the result over a photo, video or solid color, and renders the new frames to a downloadable WebM file using the browser MediaRecorder API.
Chroma key versus trained AI portrait segmentation
Green and blue screens use pixel-color thresholds, adjustable tolerance and spill suppression. The additional AI Portrait mode uses MediaPipe SelfieSegmenter to estimate a person-versus-background mask from ordinary footage. It is intended for people, not accurate separation of any object or scene. Loose hair, motion blur, occlusions, dark backgrounds, and crowded shots can still produce artifacts. Export uses MediaRecorder WebM on supported browsers; audio should be verified after download. Google MediaPipe technical documentation explains the on-device segmentation method.
Related video tools
Continue in Video Timeline Studio, Video Resizer Studio, or Video Stabilization. Technical reference: MDN Canvas Pixel Data.
FAQ
Can I remove a normal background?
For a human portrait, choose AI Portrait segmentation and load the pretrained person/background model. Normal product shots, animals and all-purpose object segmentation are not supported.
Does the tool upload my video?
No. Foreground and background media files are decoded locally in the browser.
Does it export MP4?
No. A supported desktop browser can render WebM; MP4 export is not implemented.
Remove video background: green screen vs AI portrait segmentation
For footage with a solid green or blue backdrop, choose Chroma Key mode and tune the target color, tolerance, edge softness and spill correction. For a human speaker in an ordinary room, AI Portrait mode downloads a dedicated person-versus-background model and estimates a foreground mask from video frames. The two methods solve different problems: color keying is not object detection, and the portrait model cannot perfectly remove arbitrary products, animals or all kinds of scenery.
Video preview and reliable export checks
Preview several points through the clip; hair and rapid motion often reveal masking defects. Select a 540p or 720p output on weaker devices and try Full HD on a capable desktop. Export runs through browser MediaRecorder and typically produces WebM. Check recorded duration and original audio after download. Our green-screen tutorial covers lighting, tolerance and common edge artifacts. See Google MediaPipe Image Segmenter for AI mode details.