Nvidia DLSS 5 just turned your GPU into an AI art director

Nvidia has taken the wraps off DLSS 5 at its GTC conference, and this time, the conversation is not just about performance gains. The latest iteration of its AI upscaling tech leans heavily into generative AI, with the ability to alter lighting, materials, and even character details in real time.

That shift, however, is already dividing opinions. While Nvidia positions DLSS 5 as a step towards more lifelike in-game visuals, some developers and players are questioning whether it crosses the line from enhancement into reinterpretation.

From upscaling to reimagining

Previous versions of DLSS focused on reconstructing higher-resolution images from lower-resolution inputs, using machine learning to close the gap between performance and fidelity. DLSS 5 takes a different approach. It analyses a scene’s structure and semantics, then generates new visual information on top of it.

According to Nvidia, the model understands elements like skin, hair, fabric, and lighting conditions from a single frame, then applies effects such as subsurface scattering and material-specific highlights. The result, at least in demos from titles like Starfield and Hogwarts Legacy, is a noticeable shift in how scenes are lit and textured.

It is not just environments that change. Character models appear subtly, and sometimes not so subtly, reworked. In one example, facial features look enhanced in a way that feels closer to a beauty filter than a traditional graphics upgrade. In others, lighting appears more dramatic than the scene’s original setup would suggest.

ALSO READ: NVIDIA’s new Groot N1 AI model enables robots to be more human-like

Nvidia CEO Jensen Huang has described the technology as a “GPT moment for graphics”, blending traditional rendering with generative AI. It is a bold claim, and one that reflects how far the company is pushing beyond conventional upscaling.

Control versus creative intent

The reaction from the development community has, obviously, been mixed. Some see potential in achieving more realistic visuals without the usual performance cost. Others are less convinced, arguing that generative changes risk diluting a game’s original art direction.

Nvidia maintains that DLSS 5 is not a one-size-fits-all filter. Developers are given controls over intensity, colour grading, contrast, and where the enhancements are applied. Specific objects or areas can be excluded, allowing studios to retain tighter control over how far the AI goes.

That flexibility could prove crucial. In the wrong hands, the tech risks feeling heavy-handed. Used more carefully, it may become another tool in the rendering pipeline rather than a replacement for it.

DLSS 5 is expected to roll out later this year with support for a select set of titles, including Assassin’s Creed Shadows and the Elder Scrolls IV Oblivion remake. Whether it becomes the next standard for PC graphics or remains a contentious experiment will likely depend on how developers choose to use it.

What are your thoughts about DLSS 5’s generative AI graphics upscaling abilities? Let us know in the comments below.