FreeMoCap

Research-grade motion capture normally means marker suits and expensive multi-camera rigs; FreeMoCap replaces both with ordinary webcams or action cameras and open-source pose estimation. The system records synchronized video from any number of cameras, calibrates them with a printed board, tracks full-body 3D landmarks through its skellytracker and skellycam libraries, and exports skeleton data usable in Blender, biomechanics analysis, and animation pipelines. Installation is pip install freemocap with a cuda extra for NVIDIA acceleration on Windows and Linux or a cpu build for macOS, and the GUI launches with a single freemocap command, so students and clinicians without programming backgrounds can run captures end to end. The project is AGPL-3.0 licensed with custom licensing available for other uses, has about 9,900 GitHub stars, a citable Zenodo DOI, and an active Discord community. Researchers, educators, and independent animators rely on it in settings where commercial optical mocap systems are unaffordable.