NVIDIA’s DLSS neural upscaling technology now works with virtual reality titles.
Deep Learning Super Sampling (DLSS) leverages the Tensor cores in GeForce RTX graphics cards to power a detail-enhancing neural network. In monitor-based titles, DLSS 2.0 can upscale 1080p to 4K in ‘Performance’ mode or 1440p to 4K in ‘Quality’ mode. NVIDIA claims the result is superior to native 4K rendering.
“AI upscaling” algorithms have become popular in the last few years, with some websites even letting users upload any image on their PC or phone to be upscaled. Given enough training data, they can produce a significantly more detailed output than traditional upscaling, though the algorithm is technically only “hallucinating” what it expects the missing detail should look like. In many cases there is little practical difference.
DLSS is not a driver-side feature. It requires game developers to integrate the feature into their game or engine, using NVIDIA’s provided SDK (software development kit).
DLSS 1.0 required training the network on each specific game, but DLSS 2.0 is a general solution supporting, at least in theory, any title.
Facebook is researching a similar technique for standalone VR headsets, but there’s no indication it will to ship this technology any time soon.
We aren’t aware of any VR games supporting DLSS quite yet. Most VR titles are made using Unity, which doesn’t yet support DLSS at all. We’ll keep a close eye on Unreal Engine titles.
If DLSS works as well in VR as it does on traditional display systems, it could enable some VR games to take a dramatic leap in visual fidelity on RTX graphics cards.