A Sony patent indicates that PS5 will use rescaling to improve gaming performance, a technique that, as our regular readers will know, is not only not new, but also has been key in the present generation of consoles. PS4 Pro has resorted on many occasions to a technique known as “Checkerboard Rendering”, which is characterized by rendering at a lower native resolution and rescaling until reaching the target resolution.
It may seem very simple, but the truth is that it is not. Sony has not limited itself to stretching pixels to fill gaps, that would have generated a loss of sharpness and enormous graphic quality. This technique renders only 50% of the total pixels of a scene using a 2 x 2 grid, apply a rebuild filter to add the parts of the scene that have not been processed and add temporary smoothing to limit saw teeth (aliasing).
The patent that Sony registered in 2019 starts from the same idea and seems to indicate that PS5 will use rescaling, but the technology behind that patent is completely different and is committed to an approach that reminds us, and very much, of the DLSS 2.0 system from NVIDIA, a advanced rescaling method that relies on artificial intelligence to achieve high image quality rendering only 50% of the pixels totals of each image.
PS5 will use smart rescaling – is that good news?
It is the best way to describe the technology behind that patent. Think for a moment about what NVIDIA DLSS 2.0 technology does, we have a trained algorithm to choose different images and combine them to create a single one with the best possible image quality. The result is excellent, but it represents a significant workload that is carried out through the tensor nuclei that incorporate the GeForce RTX 20 series.
The idea that Sony proposes is essentially the same, to render a game at a much lower resolution than the native one and to combine different images and frame data already used with a deep learning system to create a new image with a high level of quality. It is difficult to interpret the patent, since it is written in Japanese, but its essence is very clear, and there is no doubt that this is a technology that draws directly from the base of NVIDIA DLSS 2.0.
It is unclear how the work that would entail resorting to such smart rescaling will be performed. Sony may choose to support PS5 through the computing in the cloud, or that uses a part of the power of the GPU to this task, but in any case I find it very interesting, especially after seeing the excellent results that this type of rescaling has been able to achieve on PC.
It may be that for some the fact that PS5 will use smart rescaling may have negative connotations, but I think it is quite the opposite, is an optimization technique which has started to show tremendous value on PC, and which could also have a very positive impact on next-gen consoles.