Reproduce “Motion Diffusion Autoencoders: Enabling Attribute Manipulation in Human Motion Demonstrated on Karate Techniques” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Anthony Richardson, Felix Putze
MoDiffAE edits karate motion-capture sequences by changing a sequence-level semantic code while retaining the original stochastic code, then decoding both into motion. A linear predictor guides technique or karate-grade changes. Qualitative examples often show intended front-kick and roundhouse edits, but reverse-punch and spinning-back-kick edits are difficult. Results demonstrate a proof of concept on one imbalanced dataset, not general editing.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.