While the fields of technology and dance have historically not often intersected, recent years have seen the advent of AI-generated choreography using models trained on motion capture of a single dancer. This project will expand the state-of-the-art in this intersectional field by exploring duets featuring pairs of dancers, enabling choreography that features authentic interactions between humans & AI models.
- Extract pose information from curated videos of dance duets
- Train a GNN and/or Transformer model to analyze this data and generate new duet interaction ideas
- Create a dataset of dynamic point-cloud data corresponding to extracted motion capture poses from videos of dance duets
- Train an AI model that can generate the movements of Dancer #2 conditioned on the inputs of Dancer #1 and/or invent new, physically-plausible duet phrases
- If time permits: Learn key relationships between parts of the body of each dancer that are integral to the dynamics of the duet
- We will collaborate with the original dancers to use the model outputs to inspire new performance material
| Contributor | Approach | Repository Link | Blog Post |
|---|---|---|---|
| Luis Zerkowski | Graph Neural Network | Repo Link | Blog Post |
| Zixuan Wang | Transformer and VAE | Repo Link | Blog Post |