Nick Stracke

I'm a PhD student in the CompVis group at LMU Munich, advised by Björn Ommer. My research focuses on efficient, controllable models for visual dynamics and generative vision, including motion reasoning and generation, novel-view synthesis, and diffusion and flow models.

GitHub LinkedIn Twitter Google Scholar
Portrait of Nick Stracke

News 🔥

Research

Examples of visual concepts inferred from image sets

Show Me Examples: Inferring Visual Concepts from Image Sets

Nick Stracke*, Kolja Bauer*, Stefan Andreas Baumann, Miguel Ángel Bautista, Josh Susskind, Björn Ommer

ECCV 2026

arXiv Code (coming soon)

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

RayDer training and novel-view synthesis overview

RayDer: Scalable Self-Supervised Novel View Synthesis from Real-World Video

Ulrich Prestel*, Stefan Andreas Baumann*, Nick Stracke, Björn Ommer

ECCV 2026

Project Page arXiv Code Weights

Self-supervised novel-view synthesis that scales on abundant, dynamic real-world video.

Long-term motion embedding and kinematics generation overview

Learning Long-term Motion Embeddings for Efficient Kinematics Generation

Nick Stracke*, Kolja Bauer*, Stefan Andreas Baumann, Miguel Ángel Bautista, Josh Susskind, Björn Ommer

CVPR 2026

Project Page arXiv Code Weights

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

Probabilistic precipitation nowcasting teaser

Probabilistic Precipitation Nowcasting with Rectified Flow Transformers

Johannes Schusterbauer*, Jannik Wiese*, Nick Stracke, Timy Phan, Björn Ommer

CVPR 2026

Project Page arXiv Code Weights

Uncertainty-aware compression and rectified flow produce efficient, probabilistic precipitation forecasts.

Flow Poke Transformer motion distribution teaser

What If: Understanding Motion Through Sparse Interactions

Stefan Andreas Baumann*, Nick Stracke*, Timy Phan*, Björn Ommer

ICCV 2025

Project Page arXiv Code Weights

Multimodal motion distributions from sparse interactions provide efficient and interpretable motion reasoning.

CleanDIFT diffusion feature comparison

CleanDIFT: Diffusion Features without Noise

Nick Stracke*, Stefan Andreas Baumann*, Kolja Bauer*, Frank Fundel, Björn Ommer

CVPR 2025 Oral

Project Page arXiv Code

Better unsupervised diffusion features by removing the need to add noise.

Continuous subject-specific attribute control examples

Continuous, Subject-Specific Attribute Control in T2I Models by Identifying Semantic Directions

Stefan Andreas Baumann, Felix Krause, Michael Neumayr, Nick Stracke, Melvin Sevi, Vincent Tao Hu, Björn Ommer

CVPR 2025

Project Page arXiv Code Colab

T2I diffusion models already support fine-grained control once their semantic directions are identified.

Conditional LoRAdapter method overview

CTRLorALTer: Conditional LoRAdapter for Efficient 0-Shot Control & Altering of T2I Models

Nick Stracke, Stefan Andreas Baumann, Joshua M. Susskind, Miguel Ángel Bautista, Björn Ommer

ECCV 2024

Project Page arXiv Code

Dynamic LoRAs efficiently introduce new conditioning signals into text-to-image foundation models.

Flow-matching latent diffusion super-resolution overview

Boosting Latent Diffusion with Flow Matching

Johannes Schusterbauer*, Ming Gui*, Pingchuan Ma*, Nick Stracke, Stefan Andreas Baumann, Vincent Tao Hu, Björn Ommer

ECCV 2024 Oral

Project Page arXiv Code

Flow-matching super-resolution stages accelerate high-resolution text-to-image diffusion in latent space.