New Nintendo Patent Filed For AI Upscaling

Before I get started, I just want to put this disclaimer out there that patent filings don't necessarily mean that a company is planning to use these patents anytime soon, or at all. https://ppubs.uspto.gov/dirsearch-public/print/downloadBasicPdf/12182966?requestToken=eyJzdWIiOiJhZTNlMDM2OS02M2RjLTQ0MWItOTdkNy1mMWVhMjg0MzIwNTEiLCJ2ZXIiOiIzNjdiZmZhZC0zMDI4LTQwYjctYjM5Ni05YzNhYzJjNDNkZjciLCJleHAiOjB9

This is what I want to talk about today. Some ACTUAL, TANGIBLE, REALISTIC INFORMATION, based on a REAL PATENT filed by Nintendo themselves. This is the short-description of the patent filed on July 13th, 2023 and was released to the public on December 31st, 2024. "A computer system is provided for converting images through use of a trained neural network. A source image is divided into blocks and context data is added to each pixel block. The context blocks are split into channels and each channel from the same context block is added to the same activation matrix. The action matrix is then executed against a trained neural network to produce a changed activation matrix. The changed activation matrix is then used to generate a converted image." What all this technical jargon means is that this patent will allow for Nintendo to use some type of AI to upscale images. It doesn't say when, where, or why. It just discusses what it does. Presumably, this would be the vaunted DLSS upscaling technology that we have heard about ad-nausea to this point. You ever notice how crisp images look on small screens sometimes even on low-powered machines, but when you take that image as it is and make it larger, the image loses visual integrity? It's because only so much information is there in the smaller image. If the image is smaller, you cannot see the holes where there is no information because everything is compacted down. When you blow it up, now you can see those holes. It's just like when you take a low resolution picture and it looks worse when you try to blow it up. While this is not precisely what it is doing, in Layman's terms what the system is doing is cutting the image into small pieces, adding more visual information there to fill those holes and making each piece more dense with data, and then the AI puts the image back together again. Now, when the image is upscaled to a higher resolution, it retains the visual fidelity it did at a lower resolution.

Now, in this image here it shows how the process works in simple terms. Obviously 1080p is NOT the desired output for the Switch successor (it better NOT be.) Patents are made to be ambiguous on purpose so that people cannot get the total context on what they are planning. You'll see patents that are clearly talking about the Switch successor but they show schematics of the current Switch because they don't want you to know what the new one looks like. It could also be why Nintendo hasn't said or done anything regarding all these supposed "leaks" to the Switch successor. Maybe it doesn't matter to them because what has been "leaked" is not to hook of the system and is self-explanatory e.g. magnetic Joy-Cons, Hall-effect thumbsticks, microphone, etc. Or, what has been shown is so far off the beam that they are just letting people make complete and utter fools of themselves. They've done it before.... Getting back to the point, this technology seems to be able to double the output resolution. So, for example, if a 1080p image is fed into the system, the output image would be at 2160p which is 4K.

This image really intrigues me. Obviously the neural network is referring to machine-learning AI, but how it is being implement in this diagram is interesting. I admit that I don't fully understand what is going on here, but it seems like some type of cloud-computing. This can mean a lot of things. 1. The raw-horsepower of the Switch successor and the leaked-specs are more about "portable" mode than "docked" mode. 2. Nintendo has a habit of taking failed projects from others and making it work for them. The nVidia Shield was a failed product (more or less) from nVidia, but Nintendo used that same Tegra X1 chip for the Nintendo Switch. The motion-control mechanism in the WiiMote was turned down by numerous vendors (including Sony,) before Nintendo bought the technology and used it for the Nintendo Wii. 3. Nintendo may have a partner in this to help them cloud computing. It was kind of strange that Activision boldly proclaimed that Call of Duty games would be coming to Nintendo platforms for the next 10 years. Those are games that require a LOT of power and a LOT of storage. There would have to be some level of confidence that their games would be feasible on future Nintendo hardware to commit for that long. At the end of the day, the RUMORED codename of the console which is "Oz" I've said is possibly a reference to The Wizard of Oz film in which a scientist used his technology to make himself look larger and more powerful than he actually was. These patents seem to go along with that idea, but, when it comes to Nintendo, the only thing for sure is that nothing's for sure.