Nvidia used its SIGGRAPH 2026 keynote to show developers more control over its much-kicked DLSS 5 technology.
According to PC Games, the graphics outfit said the tool will offer different levels of controllable generation inside the system, promising “a new level of realism” without handing the whole picture to the machine.
DLSS 5 was revealed earlier in 2026 and was later described as effectively working like a 2D AI filter. However, that went down like a lead balloon. New Blood, Dusk developer, called the technology “generative AI bullsh*t”, which summed up the mood in some corners of the internet.
Now Nvidia director of applied deep learning research Edward Liu said that, with DLSS 5, a game continues “building the world exactly as the game has authored it. And the generation becomes the learned stage afterwards to enrich its appearance.”
Liu said generated images can produce different results because there is “a lot of room for the model to improvise.”
That sort of randomness is no good for real-time rendering, where the next frame cannot be picked after the fact.
“So we need some condition that is more constrained and rigorous,” Liu said.
That created three obvious headaches for Nvidia: preserving artistic intent, frame-by-frame streaming and real-time 4K.
From what Nvidia showed, developers will control how the generation appears in-game, rather than letting the AI gremlin do what it fancies. That includes options to “fine-tune intensity on a Per-Frame Basis”, covering things such as subsurface scattering, reflections and lighting.
Nvidia said DLSS 5 is not “AI replacing graphics, but AI extending graphics. The same way programmable shaders and ray tracing have extended graphics before.”
Whether players love DLSS 5 or decide it is another algorithmic coat of gloss, they will not have to wait much longer. Nvidia says DLSS 5 ships in autumn 2026.
