XeNTaX

Dead thread but:

The actual bone count is after 50 53 43 61 (PSCa). This can be compared to the parenting if this appears incorrect (further discussed below).

Parenting table starts at offset A0 where the rule is:

Parent 00 FF FF Child 00 FF FF (armature of 2 bones) (from now I will use P to refer to parent, and C for child)
When there is more then one child they are listed as: P 00 FF FF C 00 C 00 FF FF
FF FF signifies a child and the end of the armature.
When there is more then 1 bone listed between 2 sets of FF FF, the first bone chain is finished first. As far as I can tell, the chain only stops when the next number is not the previous with one added.

With pl00_ac_face the rig is:
00 00 05 00 FF FF 01 00 FF FF 02 00 03 00 04 00 FF FF
With the names at offset 1F0: AC_FACE_NULL.HEAD.FACE_EYE_R.FACE_EYE_L.FACE_JAW.(No Group)
Bones are given IDs to be used in parenting based off the order, thus giving us

AC_FACE_NULL 00
HEAD 01
FACE_EYE_R 02
FACE_EYE_L 03
FACE_JAW 04
(No Group) 05

Fitting well with the bone count, 06. If the bone count is too big to count, find the 2nd bone in parenting (No group) and add one for Null.
Thus we have
AC_FACE_NULL 00 (No Group) 00 FF FF HEAD 00 FF FF FACE_EYE_R 00 FACE_EYE_L 00 FACE_JAW 00 FF FF (please note that I made a mistake, (No group) should have no parents)

I am less sure about the positions because of the set of floats followed by 00 00 80 3F, that may have some relation to the distorted meshes, but the bone positions can be found at offset CC for the face (as floats).
This gives us the positions 0,0,0 for both AC_FACE_NULL and HEAD, and others for the eyes and jaw. (No group) appears not to have a position though, and if you compare the positions to the ones found at offset 140, you'll see only the head to jaw positions are listed, so its possible that *only* the previously mentioned vertex groups have bones.

In any case, we now have the bone names, parenting and positions. Unfortunately, the faces mesh has bones quite close together (the eyes in rest position are inside the head), however with the body the rig is a lot more clear.
(I can't seem to add another image, refer to my next post)

Here is the bodies rig (partial) (Only a section has been remade as I don't know how to code)
In relation to the possible bone co-ords (floats followed by 00 00 80 3F) 00 00 80 3F may be a scale, as within the animation files there are (if treated as a float) there are slightly different values at times.
If we consider the bone position to be the floats before 00 00 80 3F then we get some pretty interesting rigs, and they might have a connection to the distortion. (Image in next post)

This is all that I have been able to determine. Sorry for the spam, and I hope that this will provide useful to someone more capable then me!

Decided to return to this format after some time. I was able to get proper vertex positions. There is something like a bone palette near the beginning with bone ids and vertex positions. These positions are essentially same as the positions inside the vertex blocks, but they are fewer. Since the vertex count doesn't match the vertex blocks, I was also unable to use the indices. Those indices match the vertex blocks. So only point clouds for starters :



Thanks to Argg for his findings on the skeleton. Saved me some time.

I can probably make a tool for this format after a bit more research. Need to handle the indices first. I have a feeling that bone weights will be problematic too, but we'll see. I should say this is a weird format in this day and age. Spike Chunsoft is either doing something right, or something awfully wrong

Indices should be fine. If float 0 is considered to be the bone HEAD (0x3340C) (pl00_ac_face.bnc) and we consider there to be maximum 4 floats influencing a vertex at a time, the first bone weight is float 1. That leaves padding (length 4) between vertices.

Only the different positions for the vertices are listed at 0x3E0, and the count for the amount of different positions is at 0x240, followed by the face count at 0x242. Both are length 2.

Nice to see other people posting here! Good luck on the tool.

Argg wrote:and we consider there to be maximum 4 floats influencing a vertex at a time, the first bone weight is float 1.

I have tried loading those as weights, but they didn't come out great. Weird bone indices for some of the vertices. I have instead found out another weight table, starting at 0x3D90.

Argg wrote:Only the different positions for the vertices are listed at 0x3E0

They are not different positions. Same positions as the ones inside the vertex blocks. They just have a bone index at the start, which you need to transform the vertices by to get the proper world/model space vertex positions.

I am more or less done with the tool. Still I recently decided to collect my works in a single tool. I will probably release this as a module for that tool I am planning to release soon.

UPDATE

Model tool : viewtopic.php?f=16&t=19078

When I wrote the different positions I was referring to vertices with different co-ords, so there aren't any doubles, sorry. Nice work!

Kinda stuck trying to figure out this since the information is all over the place. With the padding between the bone parenting and the bone positions, is there a way to calculate the size of that or something? The size of it appears to be different between files.

Edit: Nevermind, Found that the pointer to the bone position table is actually further down in the file.

how to get textures?

I was wondering if anyone knew if this format has any kind of pointers that point towards the Faces table and Vertices? As well as maybe some kind of vertices count?

Simguy wrote: ↑Thu Jun 29, 2017 2:44 am Luckily I was able to rip some textures with a profiler, because BTX files are DDS with repeating bytes trimmed.
https://mega.nz/#!7MtjFIiD!DtLV0KJQaSy_ ... yVsQIDyppc This has a btx and a complete DDS that I ripped.
There is something going on with these that I don't fully understand, hopefully these will help someone smarter than me.

Mind telling me what you used to rip the textures?