.SkeletalMesh

[andy97]

Hi,
I've just finally extracted my desired upk (which is not supported by umodel due to its encryption). Now I'm working on .SkeletalMesh and .Texture2D. Is there any hints or tips on reading this format? I've checked umodel source code but I can't understand much about the process.

Sample: edited.

P/s: I saw that Bigchillghost had worked on this format before, hopefully you can give me some hints or tips on this format since I can't pm you.

[Pepsee]

What game is it from?

[Bigchillghost]

I've just finally extracted my desired upk (which is not supported by umodel due to its encryption). Now I'm working on .SkeletalMesh and .Texture2D.

Apparently your sample is incomplete. The vertex count of the last LOD should be 0x6E6 while there're only 0x66C vertices in the file.

Is there any hints or tips on reading this format? I've checked umodel source code but I can't understand much about the process.

I don't see it has any difference from the workflow of handling any other formats.

Weapon1_1.jpg

[andy97]

I don't see it has any difference from the workflow of handling any other formats.

Sorry for not being clear, what I meant was the structure of this file. I haven't worked on bone + skeleton much while this format begins with bone/skeleton data (or dummy data). I'm attempting to write my own script but I'm struggling to skip those data blocks to reach the mesh blocks (verts, faces, counts, etc.). And I'm wondering if all .SkeletalMesh files have the same format.

Apparently your sample is incomplete. The vertex count of the last LOD should be 0x6E6 while there're only 0x66C vertices in the file.

Thanks for the info, I didn't check it carefully. From this point, I assume that the strange block at the beginning of the file might be caused by misparsing.

Edited: I've checked it and fixed the error, the new sample is below

Sample: Here

[Bigchillghost]

I haven't worked on bone + skeleton much while this format begins with bone/skeleton data (or dummy data). I'm attempting to write my own script but I'm struggling to skip those data blocks to reach the mesh blocks (verts, faces, counts, etc.).

Nobody says you have to understand what's in these "bone/skeleton data". Just measure the distance between every FFFFFFFF marker and align them by that amount of columns then you'll have the solution figured.

Structure of your sample:

Code: Select all

... bytes ...
long	boneCount
for i = 0 < boneCount
{
	long	BoneNameIdx // refers to the string pool
	long	skip[2]
	float	rotation[4]
	float	translation[3]
	long	skip[2]
	long	Marker // 0xFFFFFFFF
} // 0x34 bytes

long	unknown
long	LodCount
for i = 0 < LodCount
{
	long	meshGroupNum
	long	Null
	for j = 0 < meshGroupNum
	{
		long	startIdx
		long	faceCount
		byte	skip[5]
	}
	byte	skip[5]
	long	totalIndexCount
	word	indices[totalIndexCount]
	
	byte	skip[0x36]
	for j = 0 < meshGroupNum
	{
		word	skip
		long	vertCount
		byte	skip[0xC]
	}
	long	totalVertCount
	long	byteIdxCnt
	byte	byteIdx[byteIdxCnt]
	byte	skip[0x20]
	float	scaling[3]
	float	translation[3]
	long	vertStructSize // 0x20
	long	totalVertCount
	for k = 0 < totalVertCount // where vertStructSize == 0x20
	{
		byte	tangent[4]
		byte	normal[4]
		byte	boneId[4]
		byte	weight[4]
		float	position[3]
		half	texcoord[2]
	}
	byte	skip[0x11]
}

And I'm wondering if all .SkeletalMesh files have the same format.

As far as I can tell, this format is similar to the one I've dealt with before, particularly you can only locate to the mesh by locating to the bone entries first.

[andy97]

Nobody says you have to understand what's in these "bone/skeleton data". Just measure the distance between every FFFFFFFF marker and align them by that amount of columns then you'll have the solution figured.

I haven't thought about that. Thanks for this tip

Structure of your sample:

Thanks, I really appreciate your help. I should try to understand bone/skel stuff as well through this structure.

[andy97]

Edited...