NVIDIA DLSS 5 and Generative AI Push Photorealism Further in Games

NVIDIA is preparing to launch DLSS 5 this fall, describing it as the company’s most significant graphics advancement since real-time ray tracing in 2018. Detailed further at SIGGRAPH 2026, the technology introduces a generative AI stage that enhances lighting, materials, shadows, reflections, and overall visual fidelity on top of a traditionally rendered game frame.

Unlike previous DLSS versions focused primarily on upscaling and frame generation, DLSS 5 uses the game’s color and motion vectors as input and applies trained AI models to enrich the image while remaining grounded in the developer’s original geometry, lighting, and artistic intent. It supports three different models with varying levels of detail and impact, along with per-object and per-scene controls that give artists fine-grained direction over the final result.

The system is designed to run in real time on a single GPU and maintain temporal stability across frames. Major publishers including Bethesda, Capcom, Ubisoft, Tencent, and others are already integrating support. Early demonstrations show clear gains in photorealism, particularly in complex materials and lighting scenarios, without requiring developers to rewrite their core rendering pipelines.

DLSS 5 positions generative AI as an extension of traditional graphics rather than a replacement, giving creators new tools while preserving the control needed for consistent visual storytelling in games.