julcs

computer graphics · neural rendering · real-time gi

scroll ↓

Julian C. Stamm
Email:moc.scluj@liamLocation:Cologne, GermanyWebsite:https://julcs.com
download cv

Education

University of Bonn

M.Sc. in Computer Science
  • Focus on Real-Time GI, Neural Rendering and Machine Learning. Expected graduation in September 2027.
Oct 2025 – present

Publications

Photon-Driven Neural Radiance Caching

Julian C. Stamm, Tom Kneiphof, Reinhard Klein
Introduces photon-based training for NRC, reducing FLIP error by up to two-thirds in tested caustic scenes compared to path-traced training. Adds around 2 ms per frame on an RTX 5070 Ti at 1440 × 1440.
May 2026
Photon-Driven Neural Radiance Caching teaser

Projects

tsnn

Slang-native neural density estimation using Slang's auto-differentiation.
  • Implemented neural spline flows, mixture models, hash grid encodings, MLPs, Adam and AdamW
  • Uses cooperative vectors and kernel fusion for training, optimization and inference
  • Achieves 1.42× the GPU inference throughput of tiny-cuda-nn JIT in image compression (512 × 512, RTX 5070 Ti)
  • Written for my lab on real-time neural path guiding [slides] [report]
May 2026 – present
tsnn neural network library teaser

Lab: Towards Real-Time Neural Path Guiding

Explored current state and real-time applicability of neural path guiding using Slang [report] [slides]
May 2026 – Sept 2026

PhotonNRC

Falcor implementation of the poster "Photon-Driven Neural Radiance Caching".
  • Implemented SPPM with hardware-accelerated inverse radius search and stochastic accumulation
  • Implemented NRC via tiny-cuda-nn and SPPC training
Feb 2025 – May 2026
PhotonNRC teaser

BiNRC

OptiX implementation of my Bachelor's thesis.
  • Built a path tracer with MIS, NEE, Disney principled BSDF, Owen-scrambled RQMC sampling and VNDF importance sampling
  • Implemented SPPM and NRC via tiny-cuda-nn with bidirectional and photon-driven training
Feb 2025 – Sept 2025
BiNRC teaser

nBounce

Personal path tracer written in Rust and wgpu.
  • SAH-based BVH construction and traversal, Owen-scrambled RQMC sampling and VNDF importance sampling
Aug 2024 – present
nBounce real-time path tracer teaser

Real-Time Glints

  • Implemented the discrete stochastic microfacet model by Deliot and Belcour in OpenGL 4.6
Oct 2023 – Mar 2024
Mesoscopic glints simulation teaser

gltemplate

OpenGL 4.6 framework developed for CG exercises.
  • Written in C++ and CMake, RAII-based resource management, texture/mesh loading utilities
Aug 2023 – present
gltemplate OpenGL framework teaser

Experience

Teaching Assistant for "Introduction to Computer Graphics"

Bonn, Germany · University of Bonn
  • Developed and maintained gltemplate, an OpenGL 4.6, C++, CMake framework with RAII-based resource management used in the exercises and animation competition
  • Held weekly tutorials on rasterization, ray tracing, BSDFs, splines, etc.
  • Assisted in creating exercises and exams
Apr 2024 – Sept 2026

Teaching Assistant for "Foundations of 4D/6D Object Capture for Virtual Environments"

Bonn, Germany · University of Bonn
  • Held daily block tutorials on NeRFs, IK, 3DGS, differentiable rendering, etc.
Feb 2026 – Mar 2026

Software Developer

Ganderkesee, Germany · Atlas Spare Parts GmbH
Developed a Python-based CAN bus logging tool on Raspberry Pi with automated email reports.
May 2021 – Sept 2021

Skills

Programming

C++, CUDA, Slang, Python, Rust, GLSL, WGSL, Java

Graphics & ML

OptiX, Falcor, OpenGL, wgpu, tiny-cuda-nn, PyTorch

Tools & Platforms

Git, GitHub Actions, CMake, Linux, Docker

Research Interests

Real-time global illumination, neural rendering, neural path guiding, neural radiance caching, photon mapping

Languages

German: Native, English: Fluent (C1; TOEFL iBT 110/120)

Others

bevy_trail

Work-in-progress bevy plugin for GPU-based trails in wgpu.
Apr 2026 – present
bevy_trail teaser

P2P Multiplayer Game

Working on a Python-moddable P2P-multiplayer card game in Rust using bevy.
Sept 2025 – present

monarch-tree

Procedural tree and butterfly-swarm animation for the course "Introduction to Computer Graphics"
  • Growing the tree with a self-devised model that steers branches using boids-like forces (up, spread, forward, noise).
Jun 2023 – Sept 2023
Monarch tree procedural simulation teaser

Tools

B&W Converter

Hardware-accelerated luminance-preserving black & white conversion.
Black & white converter teaser