VIDEO UPSCALING is hardly a new technology. From the time the first DVDs started shipping, companies have worked to create higher-quality renditions of low-resolution videos. First, it was VHS to DVD quality, then we got 720p TVs and Blu-ray drives that would turn DVD quality into Full HD quality, and now there are TVs and other devices that will upscale 1080p content to 4K, or even 8K if you have the requisite display and hardware. But upscaling video content is different from upscaling games, in a variety of ways.
For one, games have a lot more data available behind the scenes. They also run at much higher frame rates and are highly susceptible to latency issues. While the underlying concepts of upscaling may be similar, the hardware and technologies behind the scenes are much more complicated.
The proof of the pudding, as always, is in the eating—or, in this case, the rendering of the pixels. Join us as we dive into the current game upscaling solutions from Nvidia and AMD—and even Intel, sort of. Which will reign supreme, or can they all possibly coexist in an increasingly crowded world of GPUs and APIs?
THE NEED FOR UPSCALING
We’ve seen a lot of hype over the past few years around the topic of upscaling technologies. Nvidia is largely to blame, having launched DLSS—that’s Deep Learning Super Sampling to the nerds—not long after the first RTX 20-series GPUs landed in late 2018. The idea wasn’t exactly revolutionary: render fewer pixels and then intelligently upscale the result to a higher resolution. Gamers would get better performance, hopefully without losing much in the way of image quality.
The thing is, upscaling tech has been around for decades. Lanczos resampling, for example, is named after its inventor, Hungarian Cornelius Lanczos, who died in 1974. He wasn’t using his mathematical formula for video and image processing back in the day, but the principles certainly apply there. It’s also a relatively simple algorithm compared with some other alternatives, meaning it can be used without causing a
