Lukas Uzolas

Lukas Uzolas

Machine Learning · Research · Software Engineering

I obtained my PhD at CGV, TU Delft (Netherlands), with my research conducted at the intersection of Computer Vision and Computer Graphics. I explored image synthesis and 3D animation, focusing on 3D controllability to enable easy interaction with 3D representations and images.

During my PhD, I was fortunate enough to join Google Zurich (Switzerland) for a research internship. I obtained a M.Sc. degree with honours in Computer Vision and Image Processing, jointly coordinated by the Autonomous University of Madrid (Spain), University of Bordeaux (France), and Pázmány Péter Catholic University (Hungary) under the Erasmus Mundus initiative. I wrote my Master’s thesis at the Visual Computing Group at Harvard University (USA). Before this, I obtained a B.Sc. with distinction at the University of Hamburg (Germany) in Human-Computer Interaction. Next to my studies, I worked as a student software developer at various companies. For more details check my CV.

News

Rotation-Equivariant Neural Jacobian Fields for generalizable 3D mesh deformations accepted at SIGGRAPH Asia 2026.

Successfully defended my dissertation entitled Efficient 3D Deformation Learning with Neural Representations.

Surface-Aware Distilled 3D Semantic Features accepted at SIGGRAPH Asia 2025.

I joined Google Zurich as a research intern for the summer.

Selected Publications

Rotation-Equivariant Neural Jacobian Fields for Generalizable 3D Mesh Deformations

Lukas Uzolas, Ruben Wiersma, Olga Sorkine-Hornung, Petr Kellnhofer

SIGGRAPH Asia 2026

Surface-Aware Distilled 3D Semantic Features

Lukas Uzolas, Elmar Eisemann, Petr Kellnhofer

SIGGRAPH Asia 2025

MotionDreamer: Exploring Semantic Video Diffusion features for Zero-Shot 3D Mesh Animation

Lukas Uzolas, Elmar Eisemann, Petr Kellnhofer

3DV 2025

Template-free Articulated Neural Point Clouds for Reposable View Synthesis

Lukas Uzolas, Elmar Eisemann, Petr Kellnhofer

NeurIPS 2023