At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI

At this year's SIGGRAPH conference, running through Thursday, July 23, in Los Angeles, attendees can discover how leading graphics research, neural rendering, simulation and AI are transforming how worlds are created and understood by people and machines.

The NVIDIA keynote, taking place today, July 20, at 3:45 p.m. PT, will feature NVIDIA AI research and engineering leaders Neil Ashton, Edward Liu and Ming-Yu Liu discussing neural rendering techniques, world models and simulation methods for AI, built by AI.

Read on for the latest from the SIGGRAPH conference, with NVIDIA and partners showcasing:

NVIDIA keynote recap

Model Context Protocol connections bring agentic AI to content creation

New Synthetic Video Detector NIM microservice

Cosmos 3 Edge open world model for local physical AI

NVIDIA NemoClaw on DGX Station with NVIDIA Agent Toolkit

Research breakthroughs for virtual worlds and physical AI

In SIGGRAPH Keynote, NVIDIA Leaders Outline Next Era of Graphics and Physical AI Ming-Yu Liu, vice president of Cosmos Lab at NVIDIA, presents at SIGGRAPH 2026. NVIDIA AI research and engineering leaders took the stage at SIGGRAPH today to share the latest advances in neural rendering, world models and simulation. These technologies - which are changing the way digital worlds are created and used across every domain - can be applied to creative tools, industrial design, robotics and autonomous systems.

Creators and designers need tools powerful enough to expand their imagination, malleable enough to give them the freedom to shape ideas and precise enough to realize their vision exactly as intended, said NVIDIA founder and CEO Jensen Huang in an introductory video. Whether for games, cinema, robotics or factory digital twins, the goal is the same: to create virtual worlds that behave with the fidelity and realism of the physical world.

Edward Liu, director of applied deep learning research at NVIDIA, kicked off the presentation by showcasing 3D-guided neural rendering.

He shared how NVIDIA's latest advancements in neural rendering address three key research challenges: preserving artistic intent, ensuring output is temporally stable across frames and rendering 4K content in real time.

Control in artistic direction is a really exciting research direction for us, said Edward Liu. Simulation defines the world, generation enriches its appearance and artists direct the outcome AI extending graphics the same way programmable shaders and ray tracing have extended graphics before.

Neil Ashton, distinguished engineer at NVIDIA, covered innovation in AI physics with applications in weather forecasting, automotive aerodynamics, thermal design and more.

In climate and weather research, the NVIDIA Earth-2 family of open models is enabling scientists to boost their simulation resolutions to new levels with AI models trained on simulation data - while achieving accuracy equal or greater to traditional approaches.

Neil Ashton, distinguished engineer at NVIDIA, presents about NVIDIA Earth-2 at SIGGRAPH 2026. The challenge that we have is to translate the success of using these AI models in weather and climate into the world that we live in, to be able to design the planes and cars and data centers of the future, Ashton said. One of the things that NVIDIA is really focused on is helping to achieve that vision.

Ashton shared how the latest model architectures developed by NVIDIA Research and its partners compress model checkpoints down to a couple hundred megabytes, a million-times compression that enables physically accurate visualizations within a second.

Ming-Yu Liu, vice president of Cosmos Lab at NVIDIA, shared the latest advancements in world models available through NVIDIA Cosmos, a world foundation model platform for accelerating the development of physical AI systems.

What makes a foundation model a foundation model is its capability to consume enormous amounts of diverse data, he said. We trained Cosmos with data for diverse physical tasks, including world understanding, prediction, simulation and action. We use one backbone to solve all the different tasks.

Ming-Yu Liu walked through the mixture-of-transformers model architecture used to give NVIDIA Cosmos 3 global understanding. This capability enables physically accurate results that work across different embodiments - humanoid robots, grippers or self-driving cars.

Every embodiment speaks a different language, he said. Our solution is to build a common vocabulary.

NVIDIA Cosmos includes a family of open models in various sizes, supporting developers across a range of deployment use cases. Its newest addition, released today, is Cosmos 3 Edge, a 4-billion-parameter model built to run in real time, on device.

Ming-Yu Liu also announced Cosmos-Dreams, a collection of closed-loop simulators, with a colleague demonstrating a simulator built for autonomous vehicles that generates an entire world from a single frame, running on a single NVIDIA RTX PRO 6000 GPU.

Developers can use Cosmos-Dreams to verify the accuracy of their models before deploying them in a real fleet - or use it to train models on AI-generated scenarios that are difficult to create in the real world, accelerating development while saving on operational costs.

Watch the full keynote and read more about Cosmos 3 Edge.

AI Agents Expand Creative Tools to Millions Image courtesy of Epic Games. Leading creative applications are opening Model Context Protocol (MCP) connections that let AI agents work inside the tools where scenes, shots, timelines, assets and edits come to life - while creators stay in control.

For more than two decades, NVIDIA technologies - from GPU-accelerated viewports and CUDA-powered effects to NVIDIA RTX PRO ray tracing, AI denoising, neural rendering and real-time simulation - have helped accelerate the

LINK:	https://blogs.nvidia.com/blog/siggraph-news-2026/...
	See more stories from nvidia

Most recent headlines