NVIDIA Real-Time Graphics Research

Welcome to NVIDIA’s real-time graphics research group — researchers and engineers across the world working to redefine the future of photoreal, real-time, dynamic computer graphics.

Our research combines advances in light simulation, AI algorithms, and graphics systems. Our work has led to new physics-based and neural rendering technologies, generative 3D content creation, image metrics, graphics languages, and hardware features. Products derived from our inventions include DLSS, RTXDI, RTX Path Tracing, Slang, and more.

Prospective interns looking to join our team should directly contact potential research mentors, either via e-mail, social media, or in person at a conference

News

August 2025 - Our team has multiple papers at SIGGRAPH 2025:

We’re also involved in multiple SIGGRAPH 2025 courses, including:

Members

Avatar

Aaron Lefohn

Team Leader

Avatar

Andrea Weidlich

Avatar

Bart Wronski

Avatar

Benedikt Bitterli

Avatar

Chris Cummings

Avatar

Chris Wyman

Avatar

Craig Kolb

Avatar

Cyril Crassin

Avatar

Daqi Lin

Avatar

Eugene d'Eon

Avatar

Fabrice Rousselle

Avatar

Jacob Munkberg

Avatar

Jan Novák

Avatar

Jon Hasselgren

Avatar

Lifan Wu

Avatar

Marco Salvi

Avatar

Markus Kettunen

Avatar

Matt Pharr

Avatar

Miloš Hašan

Avatar

Petrik Clarberg

Avatar

Pontus Ebelin

Avatar

Ravi Ramamoorthi

Avatar

Sai Bangaru

Avatar

Simon Kallweit

Avatar

Steve Marschner

Avatar

Tizian Zeltner

Avatar

Tomas Akenine-Möller

Avatar

Tomáš Davidovič

Interns

Avatar

Alexander Root

Avatar

Blaire Yu

Avatar

Jeffrey Liu

Avatar

Kai Yan

Avatar

Kehan Xu

Avatar

Sam Estep

Avatar

Xuanda Yang

Avatar

Zheng Zeng

Publications

Quickly discover relevant content by filtering publications.

Practical Gaussian Process Implicit Surfaces with Sparse Convolutions

ACM Transactions on Graphics (SIGGRAPH Asia), 2025

doi PDF Video Supplementary Code
Practical Gaussian Process Implicit Surfaces with Sparse Convolutions

ReSTIR PG: Path Guiding with Spatiotemporally Resampled Paths

SIGGRAPH Asia (Conference Track), 2025

We present ReSTIR Path Guiding (ReSTIR-PG), a real-time method that extracts guiding distributions from resampled paths produced by ReSTIR and uses them to generate improved initial candidates for the next frame. While ReSTIR significantly reduces …
doi PDF Video
ReSTIR PG: Path Guiding with Spatiotemporally Resampled Paths

Sample Space Partitioning and Spatiotemporal Resampling for Specular Manifold Sampling

SIGGRAPH Asia (Conference Track), 2025

Caustics rendering remains a long-standing challenge in Monte Carlo rendering because high-energy specular paths occupy only a small region of path space, making them difficult to sample effectively. Recent work such as Specular Manifold Sampling …
doi PDF Video Image Viewer
Sample Space Partitioning and Spatiotemporal Resampling for Specular Manifold Sampling

An Introduction to Neural Shading

ACM SIGGRAPH Courses, 2025

Neural shading offers a new paradigm for real-time graphics, replacing hand-crafted algorithms with compact neural networks that can be trained to reproduce complex appearance. In this three-hour course, we introduce the core principles behind neural …
Videos Slides Code
An Introduction to Neural Shading

Reservoir Splatting for Temporal Path Resampling and Motion Blur

SIGGRAPH (Conference Track), 2025

Recent extensions to spatiotemporal path reuse, or ReSTIR, improve rendering efficiency in the presence of high-frequency content by augmenting path reservoirs to represent contributions over full pixel footprints. Still, if historical paths fail to …
doi PDF Supplementary Code Video Image Viewer
Reservoir Splatting for Temporal Path Resampling and Motion Blur

Unbiased Differential Visibility Using Fixed-Step Walk-on-Spherical-Caps and Closest Silhouettes

ACM Transactions on Graphics (SIGGRAPH), 2025

Computing derivatives of path integrals under evolving scene geometry is a fundamental problem in physics-based differentiable rendering, which requires differentiating discontinuities in the visibility function. Warped-area reparameterization is a …
doi PDF
Unbiased Differential Visibility Using Fixed-Step Walk-on-Spherical-Caps and Closest Silhouettes

Collaborative Texture Filtering

ACM/EG Symposium on High Performance Graphics (HPG), 2025

Recent advances in texture compression provide major improvements in compression ratios, but cannot use the GPU’s texture units for decompression and filtering. This has led to the development of stochastic texture filtering (STF) techniques to …
PDF Video1 Video2
Collaborative Texture Filtering

Generative Detail Enhancement for Physically Based Materials

ACM Transactions on Graphics (SIGGRAPH), 2025

We present a tool for enhancing the detail of physically based materials using an off-the-shelf diffusion model and inverse rendering. Our goal is to increase the visual fidelity of existing materials by adding, for instance, signs of wear, aging, …
Project doi PDF
Generative Detail Enhancement for Physically Based Materials

Radiance Surfaces: Optimizing Surface Representations with a 5D Radiance Field Loss

ACM Transactions on Graphics (SIGGRAPH), 2025

We present a fast and simple technique to convert images into a radiance surface-based scene representation. Building on existing radiance volume reconstruction algorithms, we introduce a subtle yet impactful modification of the loss function …
Project doi PDF
Radiance Surfaces: Optimizing Surface Representations with a 5D Radiance Field Loss

DiffusionRenderer: Neural Inverse and Forward Rendering with Video Diffusion Models

IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025

Understanding and modeling lighting effects are fundamental tasks in computer vision and graphics. Classic physically-based rendering (PBR) accurately simulates the light transport, but relies on precise scene representations–explicit 3D …
Project PDF
DiffusionRenderer: Neural Inverse and Forward Rendering with Video Diffusion Models
See all publications