FuzzPuppy Video Background Remover vs Robust Video Matting

FuzzPuppy Video Background Remover is a free web tool for removing video backgrounds in a browser. It uses AI removal or green-screen chroma key, then exports transparent WebM or MOV files, or MP4 with a replacement color, image, or video background. Processing happens locally in the browser; Fuzz Puppy says video is not uploaded to its servers for processing. MP4 and MOV inputs can be approximately 600 MB, with shorter, lower-resolution clips described as more reliable on modest GPUs. MODNet is the faster AI model, while BEN2 targets more detailed subjects and requires FP16 WebGPU support. AI removal requires WebGPU; green-screen removal does not. The maker lists current desktop Chrome, Edge, or Safari with WebGPU enabled for AI work. Supported mobile browsers process only the first 15 seconds at up to 720p and 24 fps with MODNet or green-screen removal. No signup or watermark is listed, and transparent MOV can be composited in compatible CapCut, Blender, and other alpha-video editors.

Robust Video Matting is a free project for separating people from video backgrounds. It processes frames with a recurrent neural network that carries temporal memory, and the project says it can perform matting in real time without additional inputs. A browser webcam demo runs the model live and can visualize its recurrent states. The conversion API accepts a video or image sequence and can produce a composited video or PNG sequence, optionally including alpha and foreground outputs.

Models are available for PyTorch, TorchHub, TorchScript, ONNX, TensorFlow, TensorFlow.js, and CoreML. MobileNetv3 is recommended for most uses; ResNet50 is larger and described as offering small performance improvements. The README reports GPU throughput figures, including 4K at 76 FPS and HD at 104 FPS on a GTX 1080 Ti, and 4K at 154 FPS and HD at 172 FPS on an RTX 3090 with FP16 under specified settings. The project cautions that file conversion runs much slower because its script does not use hardware video encoding or decoding or parallel tensor transfers. CoreML models require iOS 13 or later and do not support dynamic resolution. The code is GPL-3.0 licensed.