A recently published Nintendo patent (September 30) showcases plans for machine learning that converts resolution to a higher definition image.
Filed in March 2020 and called “systems and methods for machine learned image conversion”, the patent aims to use machine learning to convert one dataset into another using block transforms. Essentially, it would use AI to upscale the output resolution of a video game. This sounds similar to Nvidia‘s DLSS software, which currently uses AI to upscale lower-resolution inputs.
It is absolutely worth nothing that the existence of the patent is not definitive proof that Nintendo are in fact going to use this technology, just that it is an idea the company has.
The patent makes note of “image up-conversion”, which is where the resolution of an image is directly converted to a higher one. It adds that “this process can be used to show images of the first resolution on a higher resolution display. Thus, for example, a 540p image can be displayed on a 1080p television and (depending on the nature of the up-conversion process) may be shown with increased graphical fidelity as compared to if the 540p image were displayed directly with traditional (e.g., linear) upscaling on a 540 television.”
Different up-conversion techniques are also mentioned, one being that if the conversion is done in real-time during gameplay then the image quality can suffer, with the patent adding “accordingly, it will be appreciated that new and improved techniques, systems, and processes are continually sought after in these areas of technology.”
The most notable example is one very similar to how the Nintendo Switch works, where the presence of battery power changes how the machine learning upscales the image. When a system is using the battery (e.g. the Switch in handheld mode) the machine learning may not upscale the image, but that when plugged into an AC adapter it will, as the upscaling requires more power which provided by it being plugged into an AC adapter.
As the Nintendo Switch technical specs note, the video output in TV mode is up to 1080p, whilst in handheld it is up to 720p. On paper, this machine learning could allow the TV mode of something like the Switch to output to an even higher resolution than 1080p.
In adjacent news, Nintendo did recently (September 30) tweet from its corporate account stating reports of a 4K Switch were untrue, and that there are no plans for a new model of Switch.
Elsewhere, Konami has issued a public apology over the launch state of eFootball 2022, saying it will endeavor to launch an update for the game in October.
The post Nintendo patent details an AI that could upscale console resolution appeared first on NME.
