NVIDIA dedicates this week’s edition of its AI Decoded series a very direct relationship between the RTX platform and the quality of the videos we see. A very interesting topic, which we will cover below, but I also found it impossible not to relate to my colleague Isidro’s very interesting publication about the user response to the boom in processors and SoCs that integrate NPUs. A reaction which, as you can read in the mentioned article, is not very positive at the moment.

I say it’s impossible for me not to make the connection because the text, and it’s explicitly stated, of course talks about the little interest that integrated NPUs are attracting at the moment as a new paradigm of local artificial intelligence. However, the situation is very different if we talk about the interest generated by GPUs as great local AI accelerators, something that is fully accredited the weight of NVIDIA RTX graphics adapters on the marketand a wealth of AI-based features that have kept us running locally for years.

In general, when talking about artificial intelligence and the NVIDIA RTX platform, the first thing that comes to mind is a set of technologies grouped in DLSS (intelligent scaling, image generation and beam reconstruction), there are many other functions that rely on the specialization of the cores Tensor a video scaling is one of the most importantas it substantially reduces the impact of maximum bandwidth on video quality.

As we remind you in this installment of AI Decoded, upscaling is a technique that has been used for a long time, but the problem is that in general, the methods used to increase the size of an image are rather rudimentary, so the end result tends to leave a lot washing. And it is at this point, as you may have already imagined artificial intelligence translates into huge improvement compared to previous methods.

The method used by the NVIDIA RTX platform consists of take each frame and perform two actions with it. On the one hand, it performs bicubic rescaling up to the required resolution (from 1080p to 4K on the example image), on the other hand, it analyzes the image and generates the necessary elements to increase the contrast. And to this we must add that it also analyzes motion vectors in order to generate with them information that does not exist, but is coherent with the original image.

In addition, NVIDIA technology It also analyzes the received video signal in search of possible defects in it.. So, for example, if we receive a signal where some type of artifact has slipped in due to a problem in the broadcast or in the transmitter, the software will be able to identify it and discard it, instead of treating it as if it were something correct, which would cause it to be scaled .