Existing human mesh recovery (HMR) methods struggle with dynamic airborne motions, often producing unstable and unrealistic results. We present a refinement approach that enforces physical consistency by selecting reliable motion segments and applying homotopy-aware trajectory optimization. Experiments on challenging in-the-wild videos show improved realism and stability over prior methods.
@inproceedings{10.1145/3799902.3811070,author={Kim, Chanha and Won, Jungdam},title={AMOR: Airborne Motion Reconstruction via Homotopy-Aware Trajectory Optimization},year={2026},isbn={9798400725548},publisher={Association for Computing Machinery},address={New York, NY, USA},url={https://doi.org/10.1145/3799902.3811070},doi={10.1145/3799902.3811070},booktitle={Proceedings of the Special Interest Group on Computer Graphics and Interactive Techniques Conference Conference Papers},articleno={42},numpages={10},location={Los Angeles, USA},series={SIGGRAPH Conference Papers '26},}