I continue digging modelbin format
First part.
Second part.
Now i focused on 3D data parts, patient is the same AST_DB11_17 taillightL_a.modelbin.
IndB,
VerB and
Mbuf data parts have very similar headers 16 bytes in lenght, this is how they looks in decimal numbers:
Code: Select all
Name in Grub Index
(just a counter added)
Entries Count (DWORD)
Data length in bytes (DWORD)
Block size in bytes (WORD)
Units in one block? (WORD)
Unknown (DWORD)
IndB 7620 30480 4 1 42
VerB 1 2541 20328 8 1 13
VerB 2 1037 16592 16 4 37
Mbuf 1 1037 16592 16 2 10
VerB 3 12 192 16 4 37
VerB 4 472 11328 24 6 37
Mbuf 2 472 7552 16 2 10
VerB 5 986 27608 28 7 37
Mbuf 3 986 15776 16 2 10
VerB 6 34 952 28 7 37
(
better table is here)
VerB 1 have 2541 entries, and VerB 2-6 have the same amount in sum, but data blocks is different. I don't know what it means, maybe it's important...
Second half of 16 bytes long
Modl data is identical to one from body_a.modelbin, and in first part we can find a four 2-byte words, containing numbers: 46, 10, 4, 6
I think 46 it's a count of Mesh entries in this modelbin, 10 is unknown (it's 13 for body_a), 4 seems to be VLay entries count and 6 it's a count of material entries.
And finally the most interesting part -
Mesh data digging:
Some comparison pics for normal meshes and shadow ones (header part only):
This time i cut out data with additional 00 00 00 at start, for better formatting by DWORDs.
Some numbers match with ones from AST_DB11_17_taillightL_a-j.obj, this give me first clues, and i finished with this table:
(
better table is here and
latest is here)
Note that order for "Fist data byte" values is differs for LODS0 and LODs 1-5.
First mesh in table have same verts and faces count as taillightL_a_001 mesh in AST_DB11_17_taillightL_a-j.obj
Those strange hashes have matches with some of "shadow" meshes. Low detail parts share same hashes, even if other parameters differs.
I tried to find matches for lower hash in that part folder, and found it:
Then i tried in whole scene folder:
Lots of hits, but not much logic...
For detailed lod hash there's only hits with taillight
R_a.modelbin, maybe those hashes are just data size of meshes in hashed form?..
Anyway, I hope this information and table of meshes is helps to make new algo for separation of meshes in Make_H2O
![[roll]](images/smilies/icon_rolleyes.gif "rolleyes")
Working folder is uploaded
here (tables, obj's, chunks, modeldins).
.gif "Smile"){kind=small}
