Julius Erbach

I am a Ph.D. candidate at the Max Planck ETH Center for Learning Systems (CLS), supervised by Prof. Konrad Schindler (ETH Zurich), Prof. Bernt Schiele and Prof. Jan Eric Lenssen (MPI for Informatics and Saarland University). My research focuses on probabilistic generative models, such as diffusion and flow models, for computational photography and 3D modeling. In particular, I study how the priors learned by foundation models can be reused efficiently for new tasks such as inverse problems.

News

• Oct 2026 New preprint: PONDER learns in which order a masked image generator should generate, and which patches to regenerate.
• Oct 2026 New preprint: SMART, a slot memory for long-horizon video generation.
• Feb 2026 MeanFlow training paper accepted at CVPR 2026.
• Jan 2026 Continuous space-time video super-resolution with 3D Fourier Fields accepted at ICLR 2026.
• Sep 2025 FLAIR accepted at NeurIPS 2025.

Publications

PONDER: Learning to Generate in the Right Order

Julius Erbach, Dominik Narnhofer, Bernt Schiele, Jan Eric Lenssen, Konrad Schindler

arXiv 2026

arXiv Project

Long-horizon Video Generation with Slot Memory Addressed By Region Tokens

Eren Cetin, Julius Erbach, Vukasin Bozic, Hyojun Go, Zixiang Zhao, Nikolai Kalischek, Federico Tombari, Konrad Schindler, Dominik Narnhofer

arXiv 2026

arXiv Project

Understanding, Accelerating, and Improving MeanFlow Training

Jin-Young Kim, Hyojun Go, Lea Bogensperger, Julius Erbach, Nikolai Kalischek, Federico Tombari, Konrad Schindler, Dominik Narnhofer

CVPR 2026

Paper arXiv

Continuous Space-Time Video Super-Resolution with 3D Fourier Fields

Alexander Becker, Julius Erbach, Dominik Narnhofer, Konrad Schindler

ICLR 2026

Paper arXiv Project Code

Solving Inverse Problems with FLAIR

Julius Erbach, Dominik Narnhofer, Andreas Dombos, Bernt Schiele, Jan Eric Lenssen, Konrad Schindler

NeurIPS 2025

Paper arXiv Project Code Demo

EvShutter: Transforming Events for Unconstrained Rolling Shutter Correction

Julius Erbach, Stepan Tulyakov, Patricia Vitoria, Alfredo Bochicchio, Yuanyou Li

CVPR 2023

Paper

Event-based Image Deblurring with Dynamic Motion Awareness

Patricia Vitoria, Stamatios Georgoulis, Stepan Tulyakov, Alfredo Bochicchio, Julius Erbach, Yuanyou Li

ECCV Workshops 2022

Paper arXiv Code

Multi-Bracket High Dynamic Range Imaging with Event Cameras

Nico Messikommer, Stamatios Georgoulis, Daniel Gehrig, Stepan Tulyakov, Julius Erbach, Alfredo Bochicchio, Yuanyou Li, Davide Scaramuzza

CVPR Workshops (NTIRE) 2022

Paper arXiv

Time Lens: Event-based Video Frame Interpolation

Stepan Tulyakov, Daniel Gehrig, Stamatios Georgoulis, Julius Erbach, Mathias Gehrig, Yuanyou Li, Davide Scaramuzza

CVPR 2021

Paper Project Code