Uh oh... It's happened.
So, I've had a function to do this for a long time (it's how stuff like
this is possible). However, I've neglected to create an interface for it as a tool in DTW because it's kind of a Pandora's Box. I've talked to
@Achilles1515 about this before. Space can be increased for textures, palettes, subactions, animations, etc; pretty much anything. However, changing texture sizes has a relatively large impact. The game performs many operations on every pixel of every visible texture for every frame of gameplay. (Also: the GameCube's texture cache (TMEM) is only 1 MB. Not everything (6 characters, stage, items, etc.) can fit into that at once, so things must be swapped in/out as needed. More texture data would mean having to swap more often, on top of the usual graphics processing.) So if you increase the size of textures, you're incurring a slowdown to the game. The effect of one texture change would likely be unnoticeable, but with more textures changed, and larger differences in sizes between original and new, that performance impact would increase exponentially. I'm particularly concerned about console setups, and seeing costumes in tournaments.
It doesn't matter for the CSS very much, but for characters it's a whole other story, because they're loaded & rendered during a match, and therefore could affect gameplay. This opens a door of uncertainty toward character texture hacks that has never been seen before (or at least shouldn't have been). Among players, noticing this would be as subjective as perceiving monitor lag. Which means that some players will hold some distrust for character costumes; they might not be sure if they perceive a difference, and might not even be playing a character that has had any textures' sizes changed, but if they know it's possible, many will err on the side of caution and avoid them. Thus, distrust could lead to texture hacks becoming less accepted. Just like flat-screen monitors that aren't actually laggy.
There are limiting factors to how much texture data can be added, but I don't know if that would be reached before some players would notice any slowdown. Maybe a tangent, but, the largest file I've been able to load was the CSS file with 0x420000 bytes (0x430000 failed, so the limit is somewhere between those), though I don't know how much of that was texture data.
Being responsible and only doing minor texture upgrades would probably allow the game to run fine. But ATM that amount is subjective, debatable, and like I said, could lead to texture hack distrust. But maybe we can determine objective limits. Looking at the texture data in a character file to approximate the amount of processing it would require is an idea, but would only very roughly work. Because you can't simply look at the amount of texture data in a file to determine this, since some textures are used many times or in multiple places (which, yes, is able to be counted), and not all may be rendered at any given time, even if the entire character is on-screen, due to exclusion by z-order. Not only that, but even of the textures that are rendered, even if they're the same width/height/type, they do not all go through the same amount of processing, depending on a variety of factors/functions working on them. Full emulation would be required for accurate comparisons.
So, a more accurate method would be to actually perform live benchmarks. The game has native functions and performance counters for this actually, which were used by developers. Metrics are available for the graphics processor front-end, textures, vertex caches, pixels, and memory. These could be tapped into similar to other existing text-displaying hacks (like Magus' counter displays for debug mode), and displayed in real-time while you're playing the game.
Of course, even that kind of accuracy only means something if it's used. If costumes are created that can impact performance, which are uploaded and released into the wild that are not checked, then we could see an increase in people weary of playing on systems with texture hacks.
Considering that this is all just hypothesis and I haven't done any testing, I should throw in that maybe I'm just crazy. We'll need to run some tests. Commence with the science!