Nvidia’s AI-Powered Upscaling for Streaming Video Gets Mixed Reviews

Nvidia’s RTX Video Streaming Super Resolution (VSR) for GeForce RTX 40 and 30 series GPUs.

RTX VSR reduces or eliminates artifacts caused by compressing video — such as blockiness, ringing artifacts around edges, washout of high-frequency details and banding on flat areas — by using a deep learning network that analyzes the lower-resolution video frame and predicts the residual image at the target resolution. For this comparison image, standard bicubic upscaling is on the left, and VSR is on the right (Source: Nvidia)

The new set of drivers are out now, and there is some misunderstanding, and some frustration, as to what Nvidia is doing here. Firstly, the technology is very cool. Nvidia’s AI tools deliver on their promise but, they also burn up a lot of graphic processing cycles so, if you want the best image processing you have to have the best Nvidia processors. Secondly, this upscaling is solving a particular problem for browser-based streaming video, in other words, for all those people who open Netflix on Chrome or Microsoft Edge. It’s not going to turn every video on your PC hard drive into 4K. That’s two points of frustration: it doesn’t apply to all Nvidia graphics card owners, even though the problem is the same for everyone, and it’s targeted at a very specific definition of streaming video.

You can not fault people who pay hundreds and hundreds of dollars for graphics processors to want it all but, to put in perspective, AI technologies require a lot of processing power. There’s a haves and have nots split that’s going to widen as AI demand increases because, we are nowhere near having the kind of cheap processing power, or efficient algorithms, that you would need to democratize AI.