We chat about videogames and videogame accessories.

As I understand it (poorly), texture fidelity is largely an issue of system memory, which is why the Saturn port of The House Of The Dead looks so ridiculously crap compared to the arcade release. Choosing to use an anti-aliasing filter on textures appears to be a stylistic choice, especially when they are low res. That is, someone makes the decision for their textures to look "blurry" instead of "blocky" as a necessary artistic compromise.

Well, it depends on what you need out of “fidelity”.
- Texture dimensions. Larger textures (uncompressed) take up more memory, and handling a larger texture also involves more processing (not necessarily too much more depending on what you do with it, but it does) of the texels—“texture elements” (texture pixels).
- Texture appearance is a function of many things. The blurriness is one of a number of texture filtering techniques that can be chosen among (or not). Although filtering is technically a form of anti-aliasing, calling it such in a graphics discussion is liable to confuse people, because usually by that people are thinking of the concern of pixels on the display device in relation to geometry, not textures (and it can be much more computationally expensive than simple texture filtering).

The following is a grab bag of texture filtering-related terms:
- Nearest-neighbor interpolation plays a big part in kind of look you are thinking of. It is fairly simple in terms of implementation, but it results in a lot of issues in motion unless you make details so low-res it doesn’t matter. So somewhere between the motion issues and the cost of memory at the time, they made a lot of low-res textures. Me, I did not like this look as a kid but that’s really just the games I was exposed to. It’s the wrong look for realism, and you can see it in so many PlayStation games trying to do just that. It does work much better when it’s used more like a 3D kind of pixel art (maybe it should be called texel art?).
- Bilinear and trilinear filtering blurs between texels so that you don’t have a very obvious grid running across your geometry on close inspection.
- The above three filtering methods are, on a single texture application, mutually exclusive. In theory, you certainly can use more than one within a scene, but I’m have no idea whether that’s practical on real hardware, and it certainly violates normal senses of aesthetics.
- Mipmapping is the art of using smaller textures for something further away and the visual purpose that it serves is making sure things don’t look funky by picking unrepresentative texels, even blurred ones, from far away (there are other purposes this accomplishes but they are not relevant to the discussion). It is common to use with any of the above forms of filtering, otherwise you would have nasty artifacts.
- Anisotropic filtering is not the same kind of filtering as the kind of point sampling you use; rather it is more akin to mipmapping. The difference is that anisotropic filtering is more about adjusting to extreme angles so that faraway walls and floors do not feature those aforementioned nasty artifacts.

Anyway, if your hardware is not fast enough to do most the above practically (and neither the PSX nor the Saturn is), you might stretch out the texture to avoid these artifacts as much as possible (as well as save memory). One of the worst ones is definitely shimmering; when you have a faraway thing that keeps blinking in and out of existence depending on whether it happens to be in line with a pixel on the camera, it’s very distracting. Yet Another Reason™ for all that fog in early 3D games.

(and those subs are glorious I mean wow )