Nvidia has introduced a new version of DLSS 5, and it seems that Huang has achieved some sort of inner enlightenment — as if he has found a formula that combines realism and upscalers under the hood of the system.
Currently, there are only a few short demonstrations with DLSS 5, making it difficult to judge the stability and quality of the final frame — it’s more like neural rendering. In other examples, the impression is that everything appears overly lit, and the overall artistic direction of the developers gets lost. It might seem as if DLSS 5 is attempting to push games beyond the so-called “uncanny valley,” leaving them unprepared — it would have been better to preserve the original style, textures, and lighting that were initially intended.
Regarding the tech demo, two RTX 5090 graphics cards are used: one handles path tracing, while the other uses DLSS 5. Nvidia plans to combine everything into a single graphics card for mass use by the time of release.
Ultimately, we are approaching a moment when games will be built from basic primitives within the engine. This will simplify the creation of elements like AI agents. All neural rendering could operate directly on user hardware in real-time — allowing users to change skins, settings, or anything else effortlessly.
This is how technology is advancing — bringing us closer to a future where game worlds will become even more flexible and convenient for developers and players.
Created with n8n:
https://cutt.ly/n8n
Created with syllaby:
https://cutt.ly/syllaby