Kolja Bauer

I'm an ELLIS PhD student at the CompVis group @ LMU Munich, advised by Björn Ommer (LMU) and Peter Kontschieder (Meta). My research focuses on large-scale generative models and their emergent representations.

Prior to my PhD, I spent six month as an ML research intern at Torc Robotics, working with Nick Schneider and Julian Ost.

I obtained my Bachelor's and Master's degree in Computer Science with a focus on Machine Learning from KIT.

Scholar  /  LinkedIn  /  Twitter  /  Bluesky  /  CV

profile photo

Updates

  • June 2026 First-author paper on inferring visual concepts from image sets (VICIS) accepted to ECCV
  • February 2026 First-author paper on motion reasoning (ZipMo) accepted to CVPR
  • February 2025 First-author paper on noise-free diffusion features (CleanDIFT) accepted to CVPR (as an Oral)
  • October 2024 I started my PhD as an ELLIS student in the CompVis lab @ LMU Munich.

Research
VICIS hover teaser
VICIS teaser
Show Me Examples: Inferring Visual Concepts from Image Sets (VICIS)
Nick Stracke*, Kolja Bauer*, Stefan Andreas Baumann, Miguel Angel Bautista, Josh Susskind, Björn Ommer
ECCV, 2026
arXiv / Code

Inferring shared visual concepts from sets of example images and applying them to new query inputs.

Learning Long-term Motion Embeddings for Efficient Kinematics Generation hover teaser
Learning Long-term Motion Embeddings for Efficient Kinematics Generation teaser
Learning Long-term Motion Embeddings for Efficient Kinematics Generation
Nick Stracke*, Kolja Bauer*, Stefan Andreas Baumann, Miguel Angel Bautista, Josh Susskind, Björn Ommer
CVPR, 2026
Project Page / arXiv / Code

A learned long-term motion space enables efficient, goal-conditioned kinematics generation without full video synthesis.

CleanDIFT: Diffusion Features without Noise
Nick Stracke*, Stefan Andreas Baumann*, Kolja Bauer*, Frank Fundel, Björn Ommer
CVPR, 2025 (Oral)
Project Page / arXiv / Code / Twitter

Better diffusion features by eliminating the need to add noise.


This website is based on Jon Barron's website, whose code is available on GitHub.