Stefan Baumann

Stefan Andreas Baumann

I'm an ELLIS & MCML PhD student in the CompVis group, advised by Björn Ommer (LMU) and Peter Kontschieder (Meta). My research focuses on generative models for vision and their world understanding.

Prior to my PhD, I've spent a wonderful time as a ML research intern at Intel Labs (Emergent AI Research, German Ros, Benjamin Ummenhofer) and Sony (AI Speech and Sound Group, Stefan Uhlich, Giorgio Fabbro, Thomas Kemp), and as a ML researcher at a startup (Duc Tam Nguyen). I obtained my Bachelor's in EECS and my Master's in Signal Processing and Machine Learning at KIT.

I'm always open to collaborations and supervising Master's theses for exceptionally experienced and motivated students (primarily in Munich).
Just drop me an email :)

Email / LinkedIn / Scholar / X / Bsky / GitHub

Updates

All Latest

June 2026: Four papers accepted to ECCV! 🎉 Looking forward to present our works on scaling self-supervised novel view synthesis, improved discrete generation paradigms, and representation learning -- see you in Malmö! 🇸🇪
February 2026: Three papers on motion reasoning, long-term motion embeddings, and token-sparse guidance accepted to CVPR 2026! 🏔️
September 2025: One paper on disentangled motion representations accepted to NeurIPS 2025 (as a Spotlight)!
June 2025: One first-author paper on distributional motion prediction and a paper on extremely fast diffusion training accepted to ICCV 2025! 🌺
- February 2025: Two first-author papers on fine-grained diffusion control and noise-free diffusion features (as an Oral) accepted to CVPR 2025!
- December 2024: One paper accepted to AAAI 2025 as an Oral!
- July 2024: Three papers were accepted to ECCV 2024! 🎉
- May 2024: Our work on ⏳ Hourglass Diffusion Transformers was accepted to ICML 2024!
- April 2024: Neural Object Cloning, which I worked on at Intel Labs, will be used at the 2024 Olympic Games Paris!
- November 2023: I started my PhD as an ELLIS student in the CompVis lab @ LMU Munich.
- September 2023: I completed my Master's and finished my 11-month stay at Intel Labs (internship + thesis).

Research

All Selected

I'm interested in generative AI and its applications to computer vision. I also have a particular focus on motion reasoning. Selected papers are highlighted.

	What Moves? Localized Motion Representations for Compositional Scene Control Frank Fundel, Thomas Ressler-Antal, Malek Ben Alaya, Stefan Andreas Baumann, Björn Ommer ECCV*, 2026 Project Page/arXiv/Code haven't moved yet — A promptable framework for object-level motion embeddings that preserve full-scene context for localized motion understanding and transfer.
	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 / X Inferring shared visual concepts from sets of example images and applying them to new query inputs.
	RayDer: Scalable Self-Supervised Novel View Synthesis from Real-World Video Ulrich Prestel, Stefan Andreas Baumann, Nick Stracke, Björn Ommer ECCV, 2026 (Oral) Project Page / arXiv / Code / X Self-supervised novel view synthesis that scales predictably on dynamic real-world video, removing the static-scene data bottleneck.
	Envisioning the Future, One Step at a Time (Myriad) Stefan Andreas Baumann, Jannik Wiese, Tommaso Martorella, Mahdi M. Kalayeh, Björn Ommer CVPR, 2026 Project Page / arXiv / Code / X Autoregressive motion prediction from single images, 3000x faster than video world models. Enables planning via myriads of counterfactual rollouts.
	Learning Long-term Motion Embeddings for Efficient Kinematics Generation (ZipMo) Nick Stracke, Kolja Bauer, Stefan Andreas Baumann, Miguel Angel Bautista, Josh Susskind, Björn Ommer CVPR, 2026 Project Page / arXiv / Code / X Compact long-term motion embeddings for efficient goal-conditioned planning in latent space.
	Guiding Token-Sparse Diffusion Models (Sparse Guidance) Felix Krause, Stefan Andreas Baumann, Johannes Schusterbauer, Olga Grebenkova, Ming Gui, Vincent Tao Hu, Björn Ommer CVPR, 2026 Project Page / arXiv / Code Token-sparse guidance makes TREAD even better (and more efficient). Also works on MaskDiT.
	DisMo: Disentangled Motion Representations for Open-World Motion Transfer Thomas Ressler-Antal, Frank Fundel, Malek Ben Alaya, Stefan Andreas Baumann, Felix Krause, Ming Gui, Björn Ommer NeurIPS, 2025 (Spotlight) Project Page / arXiv / Code / X Abstract, appearance-disentangled representations for motion transfer & understanding.
	What If: Understanding Motion Through Sparse Interactions (Flow Poke Transformer) Stefan Andreas Baumann, Nick Stracke, Timy Phan, Björn Ommer ICCV, 2025 Project Page / arXiv / Code / X Multimodal* motion distributions from sparse interactions for efficient motion reasoning.
	TREAD: Token Routing for Efficient Architecture-agnostic Diffusion Training Felix Krause, Timy Phan, Ming Gui, Stefan Andreas Baumann, Vincent Tao Hu, Björn Ommer ICCV, 2025 Project Page / arXiv / Code / X Extremelyfastdiffusiontransformertrainingviatokenrouting. Also check out our follow-up Sparse Guidance.
	CleanDIFT: Diffusion Features without Noise Nick Stracke, Stefan Andreas Baumann, Kolja Bauer, Frank Fundel, Björn Ommer CVPR, 2025 (Oral)* Project Page / arXiv / Code / X Improving diffusion features by eliminating the need to add noise.
	Continuous, Subject-Specific Attribute Control in T2I Models by Identifying Semantic Directions (Attribute Control) Stefan Andreas Baumann, Felix Krause, Michael Neumayr, Nick Stracke, Melvin Sevi Vincent Tao Hu, Björn Ommer CVPR, 2025 Project Page / arXiv / Code / Colab / X T2I diffusion models already knew how to do fine-grained control ( ) — drag to scrub the teaser.
	DepthFM: Fast Monocular Depth Estimation with Flow Matching Ming Gui, Johannes Schusterbauer, Ulrich Prestel, Pingchuan Ma, Dmytro Kotovenko, Olga Grebenkova, Stefan Andreas Baumann, Vincent Tao Hu, Björn Ommer AAAI, 2025 (Oral) Project Page / arXiv / Code / X Efficient generative monocular depth estimation via flow matching from noisy RGB to depth.
	CTRLorALTer: Conditional LoRAdapter for Efficient 0-Shot Control & Altering of T2I Models Nick Stracke, Stefan Andreas Baumann, Joshua M Susskind, Miguel Angel Bautista, Björn Ommer ECCV, 2024 Project Page / arXiv / Code LoRAs don't have to be static! They can also introduce new conditioning into foundation models more efficiently and effectively than previous methods.
	ZigMa: Zigzag Mamba Diffusion Model Vincent Tao Hu, Stefan Andreas Baumann, Ming Gui, Olga Grebenkova, Pingchuan Ma, Johannes Schusterbauer, Björn Ommer ECCV, 2024 Project Page / arXiv / Code / X Scan order matters for SSMs in vision tasks.
	Boosting Latent Diffusion with Flow Matching (FM-Boosting) Johannes Schusterbauer, Ming Gui, Pingchuan Ma, Nick Stracke, Stefan Andreas Baumann, Vincent Tao Hu, Björn Ommer ECCV, 2024 (Oral)* Project Page / arXiv / Code Fast T2I diffusion at multi-megapixel resolutions by flowing from low to high resolution in latent space.
	Scalable High-Resolution Pixel-Space Image Synthesis with Hourglass Diffusion Transformers (HDiT) Katherine Crowson, Stefan Andreas Baumann*, Alex Birch, Tanishq Mathew Abraham, Daniel Z Kaplan, Enrico Shippole ICML, 2024 Project Page / arXiv / Code / X Efficient high-quality pixel-space diffusion at megapixel resolutions. Somehow still holding up to SOTA fidelity-wise in 2026!

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