NVIDIA updates RTX Video Super Resolution with new and more efficient AI model

NVIDIA's RTX Video Super Resolution is like DLSS for watching videos on YouTube or a streaming service like Netflix on your PC; it uses an AI model and the Tensor Core hardware on GeForce RTX cards to upscale and improve image fidelity from a lower-resolution source. For those with 4K displays, a 1080p video will become sharper with more detail restored.

Video Super Resolution (VSR), designed to remove compression artifacts and other issues from streaming videos with heavy compression, is an option all GeForce RTX owners can enable in the NVIDIA App. In addition, Video HDR uses AI to tone map SDR videos to HDR - with some awe-inspiring results.

With the launch of the GeForce RTX 50 Series and the arrival of the new GeForce RTX 5080 and RTX 5090, NVIDIA has brought several new features to gamers - from DLSS 4 to NVIDIA Broadcast updates for creators. As part of these updates, NVIDIA has confirmed that RTX Video Super Resolution is running on a more efficient AI model.

One that will use up to 30% fewer GPU resources when using the highest quality setting. According to NVIDIA, this will allow more GPUs, especially Turing-based GeForce RTX 20 Series cards, to enable VSR with the highest quality setting.

With this update, VSR can also upscale HDR video and give everyone more control over Video Super Resolution; there's a new option to set GPU Utilization. At the 'High' setting, VSR will use as much GPU horsepower as possible to deliver the best image quality.