NVIDIA DLSS 5 Aims for Photorealism with AI Rendering

Daily Technology

17/03/2026

NVIDIA has announced its next-generation upscaling technology, DLSS 5, positioning it as a significant leap forward in computer graphics. Unveiled just months after DLSS 4.5, this new iteration heavily integrates artificial intelligence, with the company claiming it can infuse game visuals with photorealistic lighting and materials using a real-time neural rendering model. The technology is slated for release this fall.

The Core Technology

DLSS 5 operates by taking a game's color and motion vectors for each frame as inputs. An AI model then processes this data to generate and apply photorealistic lighting and material properties that are anchored to the source 3D content. According to NVIDIA, this process ensures visual consistency from one frame to the next. The entire rendering pipeline runs in real time and supports resolutions up to 4K, aiming to deliver a new level of visual fidelity without the traditional performance overhead.

Performance and Demonstration

During a keynote demonstration, NVIDIA showcased DLSS 5 with popular titles such as Resident Evil: Requiem, Hogwarts Legacy, and Starfield. The visuals exhibited a noticeable increase in detail, particularly in character hair and skin tones. However, these demonstrations compared scenes with DLSS 5 enabled against the same games running with no DLSS features active. This makes it difficult to objectively assess the incremental improvement over its immediate predecessor, DLSS 4.5, especially when combined with advanced features like path tracing.

The demonstration system was powered by two RTX 5090 GPUs, though NVIDIA stated that the technology will eventually be optimized to run on a single, powerful graphics card. The company frames DLSS 5 as a step toward achieving Hollywood-level visual quality in real-time rendering, effectively providing developers with a controllable generative AI video model.

Market Context and Comparison

NVIDIA has labeled DLSS 5 the "biggest breakthrough in computer graphics" since the introduction of real-time ray tracing in 2018. While DLSS 5 promises to generate photorealistic pixels through AI, its market adoption remains a key question. For comparison, real-time ray tracing, despite its visual benefits, has not yet become a mainstream feature for all gamers due to its demanding hardware requirements.

DLSS 5 represents a different approach from previous versions, which primarily focused on upscaling lower-resolution frames to boost performance. This new version shifts the focus to generative AI for enhancing material and lighting quality directly. The success of DLSS 5 will depend on its performance on single-GPU systems and whether the visual enhancements are compelling enough for widespread developer and player adoption.

2026-03-09

Understanding the Next Wave of AI with GPT-5.4

Explore the key advancements in OpenAI's new GPT-5.4 model, from powerful agentic AI capabilities to significant improvements in accuracy and reliability.