I still think of it as an upscaling method though, and I think it's silly to argue that it isn't, as opposed to making sure everyone understands it's NOT traditional upscaling, something which I don't think anyene in here thinks. But I guess we are left arguing semantics.
Semantics matter. We don't call all four-legged domesticated carnivores "dogs", because it matters that some of them are cats. Having similarities doesn't make things identical. And in the case of checkerboard rendering, the similarities to upscaling are swamped by the differences.
It's fancy upscaling or upscaling with a different algorithm.
No it isn't. Traditional rendering derives pixel values from knowledge of 3D objects' position, motion, and orientation, augmented by other data, including the values of adjacent pixels.
Checkerboard rendering derives pixel values from knowledge of 3D objects' position, motion, and orientation, augmented by other data, primarily the values of adjacent pixels.
Scaling derives pixel values from the values of adjacent pixels, sometimes augmented by other pixel values.
I think this makes it very clear that checkerboard is a type of rendering, not a type of scaling.
Data is being generated where it isn't, whether you do that by duplicating or filling the gaps on a deferred context it doesn't change the point of the technique.
All rendering generates data where it isn't, but it'd be ridiculous to call it all "scaling".
We have up scaling and down scaling to make images bigger or smaller. How about we call checkerboarding something like "lateral scaling?"
Because it isn't scaling, it uses a very different method far more akin to traditional rendering.