And I'd disagree about the spectrum analysis as it detects only one part of the problem. Keeping the native high frequency details is also important. With VRS you degrade the original assets and you get those blocky textures everywhere (you also get some ugly artifacts on reflections in Doom Eternal BTW).
I think at the heart of the debate perhaps, but as you can see with different upscaling algorithms using a different base resolution it’s able to separate the blockier from the finer.
Essentially, a quick method to determine if an image has more detail or less detail is to count the number of unique pixel colour values it has. And I’m theory if every colour value is different you aren’t really getting this colour blocking.
Spectrum is sort of similar, except it does it by frequency, what are the frequencies that need to be added together to form the image. The more finer the detail the more frequencies that need to be added. And then we look at the total frequency and see which ones have the most.
So in some ways it accounts for blocks, but in other ways it does not because it’s a summation of the total image and not a particular spot you’re looking at.
As per why to me it makes sense to degrade detail, it just comes down to persistence of motion, which is why we can see animation in the first place.
Those small details during motion are going to be gone for several reasons, most of which is where your eyes are actually focused on the screen, and secondly when things are moving your eyes can’t possibly see that level of detail the faster things move.
So I guess in my preference the areas that are going to be in focus or visible should be sharper, because that would actually be more in line how we see things, we aren’t able to actually stop our vision and look at all the blurry stuff that was out of focus.
I don’t see a problem with VRS when done correctly especially as we move to foveated rendering, we do have an evaluation problem around VRS.