SIGGRAPH returns to Los Angeles from July 19 to 23, bringing together researchers, artists, game developers, hardware makers and visual effects specialists for another look at where computer graphics is heading next. This year, the answer appears to be deeper into AI, neural rendering, simulation and interactive systems, as technologies once confined to research papers move closer to production use.
NVIDIA will use its sponsored keynote, “Next Era of Graphics — Neural Rendering, World Models, and Simulation,” to outline how machine learning is reshaping digital content creation. Researchers Jan Kautz, Ming-Yu Liu and Edward Liu are expected to cover neural rendering, world models and AI-driven simulation, supported by live demonstrations.
This year, SIGGRAPH 2026 will also introduce its first dedicated Games Summit, a one-day event covering accessibility, audio, performance capture, production technology and the increasingly difficult economics of game development. The program is intended to bring together developers, researchers and technology suppliers while acknowledging that impressive graphics are of limited use if studios cannot afford to finish the game before the money or staff runs out.
Away from the formal sessions, the Experience Hall will once again provide the conference’s more experimental side, combining spatial storytelling, immersive technology, digital art and emerging interfaces. Among the exhibits is “Dog Walk,” which features two robot dogs whose onboard language models develop through interaction with their human co-parents. The project explores companionship, AI alignment and the apparently inevitable human tendency to become emotionally attached to machines with convincing eyes.
Other installations include “Flashover,” which uses volumetric imagery to recreate memories of wildfire, and “Ancestral Craft and Emerging Technologies,” which combines extended reality and robotics with knowledge developed in the Amazon rainforest. The Immersive Pavilion will feature augmented reality, eye tracking, biofeedback, multisensory interfaces and AI agents. Projects include “Landmine/Garden,” which places visitors in a simulated minefield, and “Impulse,” which combines motion capture with Unreal Engine to let users choreograph digital performers in real time.
There is also “Smileoyed,” a virtual pet built into a cushion-like device that reacts to breathing and facial expressions, along with “Goo Goo Clash,” which turns the classic game Snake into a co-located mixed-reality sport because apparently moving a few pixels around a screen was no longer demanding enough.
AI will dominate much of the conference, but the more important developments are likely to come from hybrid workflows rather than fully automated content creation. Neural rendering could reduce the brute-force cost of complex imagery, world models could make simulated environments more responsive, and AI-assisted tools could allow smaller studios to produce work that once required much larger teams.
Whether that makes production more efficient or simply encourages companies to demand twice as much work in half the time remains to be seen.
SIGGRAPH has always offered an early look at technologies before the rest of the industry catches up. This year’s event suggests the next generation of graphics will not just render digital worlds, but generate, simulate and respond to them. Just as in the years of past, the pixels are still there. This year they are simply starting to think for themselves.
