Portfolio item number 1
Short description of portfolio item number 1
Posts by Collection
portfolio
Portfolio item number 2
Short description of portfolio item number 2
publications
Frustum Volume Caching for Accelerated NeRF Rendering
Published in Proceedings of the ACM on Computer Graphics and Interactive Techniques, 2009
Speeds up NeRF rendering by caching non view dependent MLP features in a froxel grid.
Recommended citation: Michael Steiner, Thomas Köhler, Lukas Radl, Markus Steinberger (2024). "Frustum Volume Caching for Accelerated NeRF Rendering." PACMGIT. 7(3).
Download Paper
Image-Based Spatio-Temporal Upsampling for Split Rendering
Published in Computer Graphics Forum, 2025
Enables smooth, high-quality rendering on ultra low-powered mobile devices by generating intermediate frames entirely from images using a novel spatio-temporal interpolation split rendering pipeline—without needing geometry or real-time rasterization
Recommended citation: Thomas Köhler, Michael Steiner, Lukas Radl, Brian Budge, Markus Steinberger (2025). "Image-Based Spatio-Temporal Upsampling for Split Rendering." Computer Graphics Forum. 8(44).
Download Paper
LoD of Gaussians: Unified Training and Rendering for Ultra-Large Scale Reconstruction with External Memory
Published in ArXiv, 2025
Enables abitrary scene scale for Gaussian Splatting reconstruction on single consumer GPU.
Recommended citation: Felix Windisch, Thomas Köhler, Lukas Radl, Michael Steiner, Dieter Schmalstieg, Markus Steinberger (2025). "LoD of Gaussians: Unified Training and Rendering for Ultra-Large Scale Reconstruction with External Memory." ArXiv.
Download Paper
NeuralPVS: Learned Estimation of Potentially Visible Sets
Published in SIGGRAPH Asia, 2025
Provides state of the art from region PVS calculation using deep neural networks.
Recommended citation: Xiangyu Wang, Thomas Köhler, Jun Lin Qiu, Shohei Mori, Markus Steinberger, Dieter Schmalstieg (2025). "NeuralPVS: Learned Estimation of Potentially Visible Sets." SIGGRAPH Asia.
Download Paper
AAA-Gaussians: Anti-Aliased and Artifact-Free 3D Gaussian Rendering
Published in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2025
Highlight
Introduces a fully 3D evaluation and filtering pipeline for Gaussian Splatting that eliminates aliasing, projection distortions, and popping artifacts through adaptive 3D smoothing, stable view-space bounding, and efficient 3D culling, enabling real-time artifact-free rendering even for out-of-distribution views.
Recommended citation: Thomas Köhler, Michael Steiner, Lukas Radl, Felix Windisch, Dieter Schmalstieg, Markus Steinberger (2025). "AAA-Gaussians: Anti-Aliased and Artifact-Free 3D Gaussian Rendering." ICCV.
Download Paper
talks
Image-Based Spatio-Temporal Upsampling for Split Rendering
Published:
I presented our paper “Image-Based Spatio-Temporal Upsampling for Split Rendering” at the High-Performance-Graphics conference in Denmark 2025.
Rendering and Geometry from 3D Gaussian Splatting
Published:
Bernhard Kerbel, Felix Windisch and me presented current state of Gaussian Splatting, the artifacts that can occur during rendering and how to avoid them as well as how Gaussian Splats can be converted to meshes.