I'm a senior researcher and computer science lecturer at the University of Cologne, specializing in high performance computing, computer graphics and 3D scientific visualization. My research is centered around real-time ray tracing, a technique to create virtual images from 3D data aiming at high realism and the simulation of light transport in a physically inspired way. Here's a bunch of images that involve ray tracing to some extent and that I used in my projects and publications. I'm also the author of a number of open source projects that have to do with ray tracing and direct volume rendering. Find more details about those on the Projects page on this website. And I'm randomly blogging about my research topics under http://blog.szellmann.de.

I'm on the international program committee of the IEEE VIS 2021 conference and was a regular reviewer and IPC member at the Eurographics Symposium for Parallel Graphics and Visualization (EGPGV): committee member in 2020 and 2021, reviewing in 2019.

In addition, I also regularly review papers for the IEEE VR conferences, for IEEE Access, and for several smaller international journals and conferences.

I'm currently teaching the master students' lecture Architecture and Programming Models for GPUs and Coprocessors (APGC) at the University of Cologne:
2022 Lecture (redirects to the University of Cologne).

I designed this lecture from scratch in 2018 and am now holding it for the 4th time. In a regular semester, it is comprised of 2x90 minutes lecture + 90 minutes exercises per week ("6 SWS").
The main topics are:

• Processor architecture.
• Parallel algorithms and parallel programming.
• Computer graphics (rasterization, ray tracing).
• Graphics pipeline in hardware.
• GPGPU programming.

Here's a link to a youtube channel with screen recordings from 2021:
https://www.youtube.com/channel/UCH57k49l-E_nqkaXH4HZQug.

Direct link to the lecture's playlist on youtube:
https://www.youtube.com/watch?v=Ehaw8Xo0pwY&list=PL8GFXx31a3wrgwjvi2FIS4-b95XFevAyj.

More material and links to past semesters can be found on the institute's websites: 2021, 2019 and 2018.

I've also regularly supervised seminars on computer graphics and GPGPU programming, as well as masters / diploma theses on topics such as FPGA programming, volume rendering and empty space skipping, reconstruction filters, etc.

My CV as of 3/3/2021 (direct link).

As a researcher I work on a couple of (mostly software) projects that are related to 3D rendering, high performance computing, real-time scientific visualization etc. The most notable projects I'm currently collaborating on are:

I was able to raise external funding with the DFG (project no 456842964) to support my research on time-varying AMR visualization. I'm planning to start working on this at some point in 2022. More information to follow. The work will be based off of the ExaBrick paper (see below, and also this external website at the UoC:
https://vis.uni-koeln.de/forschung/software-exabrick-amr-rendering-framework).

That's a bunch of projects I collaborate on with researchers from SCI (Scientific Computing and Imaging Insitute, University of Utah) and from NVIDIA. OWL is the "OptiX 7 Wrappers Library" (see this blog post by Ingo Wald). Some notable projects where we used OWL to make use of hardware-accelerated ray tracing are:


ExaBrick is an optimized AMR (adaptive mesh refinement) visualization data structured for large simulation data that is optimized for NVIDIA GPUs with hardware ray tracing cores. See our paper: I. Wald, S. Zellmann, W. Usher, N. Morrical, U. Lang, V. Pascucci (2020), "Ray Tracing Structured AMR Data Using ExaBricks" at IEEE VIS, as well as the software repository: https://github.com/owl-project/owlExaBrick


That was a paper project for High Performance Graphics 2020, where we used OptiX hardware instancing to trick the ray tracing BVH into performing an OBB (oriented bounding box) test to safe on context switches between RT cores and shading units.


Here we used ray tracing hardware to perform radius searh queries in large 2D graphs to compute layouts using the Fruchterman-Reingold algorithm. That project culminated in the VIS 2020 short paper: S. Zellmann, M. Weier, I. Wald (2020), "Accelerating Force-Directed Graph Drawing with RT Cores". Also check out my blog post on this algorithm: http://blog.szellmann.de/2020/10/17/graph-drawing-with-rtx/

That's an attempt to provide an implementation of the OWL interface on the CPU. Very experimental, source code can be found under: https://github.com/szellmann/fakeOwl

Other projects are focused on glyph visualization or on large data rendering. For more details, check out the "OWL" project website and the blog post mentioned above.

Visionaray is a C++ template library providing a bunch of different algorithms and data structures related to ray tracing that are highly optimized for x86 and ARM CPUs as well as NVIDIA GPUs. Here are some images that were rendered with Visionaray. Click to enlarge (opens a new tab).

The most important Visionaray-related paper is: S. Zellmann, D. Wickeroth, U. Lang (2017), "Visionaray: A Cross-Platform Ray Tracing Template Library", which was presented at SEARIS, a workshop at IEEE VR 2017. Visionaray was used in numerous different projects, for example the instancing / large 3D model project from EGPGV 2020: https://github.com/ukoeln-vis/instanceviewer or the VMV 2017 project on compile time polymorphic ray tracing: https://github.com/ukoeln-vis/ctpperf.

Volkit is a relatively new library (still under heavy development) implementing algorithms centered around 3D volumetric data. More to come very soon!

The following is a relatively well-maintained list of my peer-reviewed scientific publications. Also see my Google Scholar profile or my profile on ResearchGate. A link to my PhD thesis on "High Performance Volume Rendering" (Cologne 2014, primary advisor: Prof. Ulrich Lang) can be found here: https://kups.ub.uni-koeln.de/5727/.