Research

Papers

Work on learned structure for graphics, vision, and generative world models. Links go to the paper, project page, and code when available.

Peer reviewed

Publications

Neural scene representations

Reusable neural primitives and lightweight instances for compact, editable 3D scenes.

William Koch, Amogh Joshi, Cyrus Vachha, Cheng Zheng, Felix Heide

Generative world models

Vision-language controllable, vectorized latent scenario generation.

Lili Gao*, Yanbo Xu*, William Koch*, and collaborators

Geometric learning

Learning graph-based motion hierarchies directly from observed dynamics.

William Koch*, Cheng Zheng*, Baiang Li, Felix Heide

Preprints & theses

Preprint · MSc thesis · University of Oxford

Efficient novel-view synthesis for moving faces from ordinary webcam input.

Preprint · BSc thesis · École Polytechnique

A secure multi-party framework that prevents unilateral tracking.