Sony appears to be working on bringing real-time image upscaling to PS5, and potentially PSVR PS5.
A new patent, filed by Sony Interactive Entertainment (via Resetera) outlines a plan for a "computer-implemented method for completing an image." Traditionally, images can contain "regions of missing or corrupted image data." Various processes exist for filling in those gaps, but machine learning techniques can be performance-intensive or time consuming.
To bypass those issues, Sony's new computer-implemented method involves generating a 'mask' of the image, revealing some sections of it and hiding others, before altering it, and applying filling processes based on the presence or absence of holes. Eventually, the computer will learn a relatively simple means of categorising whether a hole should be there or not, allowing it to fill things in more quickly, with less computational power.
It's not too dissimilar to Nvidia's Deep Learning Super Sampling (DLSS) tech, which uses upscaling technology in its graphics cards to improve image resolution in real time. Sony, however, seems to be working on bringing its version of the tech to PS5, and potentially PSVR PS5.
The patent notes that the system could be used within a virtual reality device, noting that "Virtual reality headsets require much higher computing power to display a satisfactory image to a user than a conventional computer monitor," noting the need to use less computing power to aid with "user comfort and immersion."
The LinkedIn profile of Andy Bigos, senior principal engineer at Sony, who is listed as one of the patent's inventors, states that he has recently been "working on the application of machine learning and neural rendering to real-time graphics on the PlayStation 5." That doesn't specifically mention VR, but does seem to link this tech directly to the console.
This new tech could get these upcoming PS5 games looking even better.