just as a test I used my old forza3 app (from 4 years ago) and was surprised, it worked (with unplaced meshes, though):
.

I really just go by the patterns of data from Model Researcher which mariokart64 said was a bad thing to do. Depends on what you mean by "the patterns of data". Using search-patterns in a hex editor to get "an idea of the data structure" is not that bad, if you have experience with some dozens of 3...

Interesting format, guess it's using child objects (I didn't care for).
Here's the 3 biggest meshes I could get:
.

Sharppy wrote: ↑

Fri Apr 09, 2021 1:09 am

The data seems pretty clear.

Really? Tell us more.
How to get the different FVFsizes (84, 36, ?).

How to get the structure (hierarchy, if any)?

But its very time consuming. Is there anyway to parse the model data by a script ? I wouldn't know, why not - but it's also very time consuming. :D Counts are to be found in the ..._unpacked files, that's good. Get the start_of_face_indices_blocks by searching for 0000 0001 0002 in 0xf9785b04_unpac...

by the way why in one model FVFsize is 6 and in another is 12, how would I know in what model is what value? :( Half floats, 2 bytes, so 3 x2= 6. Floats, 4 bytes, so 3 x4=12. Find the floats (learn to recognize them, read the tutorial, "tut" button in hex2obj ). If there are no floats then try half...

r7037-nif-uvs.png . 0x5F2E is the start of a new block/a signature part, whatever - simple to see with some experience (0x5F3B: start of face index block). 4 is length of two half float uvs, so start address of uvs' block is 0x5F2E - 1735x4 (hex.)= 0x4412 (This calculation is different for nif r289...

but it is still difficult for me to find the exact values of the UV block, what would be the value of the FVF size of the nif 289? dunno what you mean, it's 12, as used by you. Or speaking of "size UVB" (size of UV block)? I tried with 1E168 there's floats, not half floats, you won't get far as lon...

Here you go:
. (The uvs look a little bit strange - I tested a sub mesh, one of those center circle objects, and they looked ok to me.)

yeah, it's a matter of finding the suiting face indices blocks:
.

Without a map sample no one can tell, I guess.

Magilou wrote: ↑

Mon Apr 05, 2021 10:08 pm

Does that mean it can't be done?

At least offzip did not work (use forum search to understand how it works).

(So you'll need someone to uncompress the .xb files.)

Hello, ... They're using the same .xb files Hello, why do you think so? In fact they are different, compare any of the EGWT xb files to a HSGW xb: . HSGW_PS3_xb_header.png The "29000000" is a count (41 decimal for said model, used in for m in range (A[1]):) For most EGWT xb it's 00000003 (50331648 ...

uvs are hard to track, here's some of the easier ones:
.

moonpaladin wrote: ↑

Sun Apr 04, 2021 6:14 am

I'm using 1F67E for star address but it fails,

Didn't care for uvs but what is the logic behind using (0x)1F67E?
Scrolling down in the file after that FF.. block there's only ONE possible address for me. Much to learn you have... use the source!
.

Thanks - well, did you read here? Maybe I was not clear enough: Make_obj Skyforge version DOESN'T work with those samples, because their first face indices blocks doesn't start with 000001000200. You need to be a coder to adjust the code or simply use hex2obj . But, of course, there is a way, which ...