A new patent from Sony regarding image reconstruction technology in the vein of NVIDIA's Deep Learning Super Sampling technology that could be used for the upcoming PlayStation 5 console, surfaced online recently.
The new patent, which was filed back in 2019 but was only published on July 23rd, details technology that is capable of reconstructing an image using a plurality of reference images through machine learning.
The patent is sadly only available in Japanese as of now, so many of the details are still unknown, but from the general description, it seems like this system will be an improvement over Checkerboard Rendering that is used by the PlayStation 4 Pro console.
An information processing device for acquiring a plurality of reference images obtained by imaging an object that is to be reproduced, acquiring a plurality of converted images obtained by enlarging or shrinking each of the plurality of reference images, executing machine learning using a plurality of images to be learned, as teaching data, that include the plurality of converted images, and generating pre-learned data that is used for generating a reproduction image that represents the appearance of the object.
As companies do file patents for technology that eventually goes unused, it is not certain that the PlayStation 5 will feature this image reconstruction tech that is detailed in the new Sony patent. NVIDIA's DLSS 2.0 technology is proving how such technology can result in better visual quality at lower resolutions, so there's no doubt that something similar would benefit next-gen console games greatly.
The PlayStation 5 launches later this year worldwide. A precise release date, as well as final pricing information for both the standard edition and the Digital Edition, has yet to be confirmed. We will keep you updated on the console as soon as any announcement is made, so stay tuned for all the latest news.
