Alrighty, not sure how everyone's going to like this but we have what appears to be a working model for how texture pages are decided in the dual-textured models. Let's take a look at two quads from Dario's UV map. Although Dario is a human-size character, his model has enough detail to require two textures.
Quad Example 1:
000011D0 - 5B F3 57 F3 5C F1 57 F1
000011E0 F0 04 00 05 F8 01 18 02
Quad Example 2:
000011F0 - 36 EB 2F F5 2E E3 28 EC
00001200 81 12 21 13 88 09 D8 09
The difference between these two quads is that the first, which has vertex index pointers with run-of-the-mill values, is mapped to the LEFT texture, whereas the second, which has 1s instead of 0s at the end of the first two vertex pointers, is mapped to the RIGHT texture.
Thus, an extra rule for quads:
If the first two vertex pointers end in a "1" instead of a "0" nybble, then the texture piece is pulled from the RIGHT texture.
If the first two vertex pointers end in a "0" instead of a "1" nybble, then the texture piece is pulled from the LEFT texture.
Luckily I've not seen any situation in which either of the first two vertex pointers ends in an "8" nybble, or we'd be in trouble.
The rule is the exact same for triangles, but keep in mind that there's a switcharoo in which the last two vertex index pointers go first and the first goes third. Thus, we tell the difference between a LEFT triangle piece and a RIGHT triangle piece as follows:
Triangle Example 1:
000006E0 15 23 11 0D 19 11 E0 03-E1 15 D1 07
Triangle Example 2:
000006F0 - 47 1E 39 1B 39 21 50 06
00000700 40 09 40 0C
So the game engine reads the first two pointers first, then the third, or at least that's how I think of it. Thus, the top triangle is mapped from the RIGHT texture, and the bottom triangle is mapped from the LEFT texture.
Consequently, we should not see any "8" nybbles in the second and third vertex pointer slots of triangle pieces. I have never seen a "9" nybble where an "8" nybble should be, so I'm hoping the modeling program the developers used was careful where it placed its vertex indices.
I'm a bit miffed that a texture page indicator would be stuffed into vertex pointers, but it's an accurate indicator of which texture the game engine is looking at when it pulls UV pieces based on manual UV mapping exercises. Halkun, Cyb, and M, does this seem plausible? Luminaire, is this something that can be factored into the Blender script?
Also Luminaire, Satoh and I have been keeping an eye on the dimensions of the weapon, field model, and battle model textures we've been coming across. It appears that powers of two is the way to go. The weapon textures, for example, have dimensions of 32 x 128 if I remember Satoh's reports on those correctly.
Attached is an Excel spreadsheet that segregates Dario's UV map into triangle and quad sections based on the constructs and Dario's model data in the zip. The first texture attached is his LEFT texture, and the second texture attached is his RIGHT texture in case anyone wants to rummage through the data and see how this works firsthand.
[attachment deleted by admin]