made some progress: http://www11.pic-upload.de/thumb/24.04.15/d1kexifvkoof.jpg Attached some H2O files. You'll need hex2obj_023 e .exe to create obj files from them. (Archive hex2obj_0.22.zip required, too). Load Bike00 1 .bin into hex2obj then one of the H2O files. Press the ' mesh ' button. BIKE00...

dunno, You might read the Wild Arms 3 thread where meshes were extracted correctly (more or less).
Maybe it's similar.

RiccoChicco wrote:[...] and extract all customisable parts somewhere else.

to extract all the parts is the problem atm: the ones left from the black line are crappy and a whole bunch seems to be missing:

Is "shortall signed" the same as 2ByteSigned? should be. The scaling/dividing factor may be different. Also I don't understand how someimtes you get meshes that look like nothing (in your first point), but other times look like something like in your last post. Don't understand it, too. H...

whoops. What a tedious task: And that's the wheels only, different types, lods, whatever.

I didn't say that it's not signed short. As you can see the above point cloud is drawn using them.

Here's another promissing structure using "ShortAll" (signed):

I made a cleaner version of the script with real variable names yeah, that's better (and bigger :D ) thx. The question is what format are the vertexes in, because assuming 2byteSigned vertexes and converting them to floats only yields unrecognizable objects. yep. This point cloud looks interesting,...

yeah, is very likely to become my next Make_H2O project! (If not anyone else will release a complete max, Noesis, whatever script... :D ) chassis#chassis_0@Bike001 Bike001.JPG searching for the tables containing counts and offsets (if any). Feel free to contribute otherwise it may take some time... ...

I'm having a hard time, figuring out the format of the vertexes. I'm now thinking that they might be half-floats instead of signed integers, what do you guys think? I've left the path of wild guessing... I'd a look at the 5.3ds file which was created by the bms script. Vertices chunk id (4110) and ...

update

114 back hair does not appear found it in the second "KFMG" section: 0x20da70 1170 Vb1 76 40 0x212940 713 021000 0x0 255 Updated Make_H2O.exe. Use it with hex2obj0.23 c .exe! I didn't fully understand the counts in the tables and did some workaround to get a maximum count of obj files. Th...

Look forward to your correction for Dl0114, hair, use *_0.H2O: 0x12240 6277 Vb1 76 40 0x2a400 3057 021000 0x0 255 and new two error 105 for DL0105 use *_9.H2O (right leg) 0x168a0 1130 Vb1 76 40 0x7010c 554 021000 0x0 255 *_12.H2O (front hair) 0x17f18 342 Vb1 76 40 0x8ae54 201 021000 0x0 255 108 err...

using hex2obj (view link in my sig) for a quick format scan:

Thank!! but hair error In the DL0114*_0.H2O change the vertex count from 4801 to 3057 (as supposed by hex2obj when clicking the go1 button) 0x12240 9478 Vb1 76 40 0x2a400 3057 021000 0x0 255 Sadly there will be superfluous triangles. Dunno why. ------------- For DL0124 try the appended H2O files. (...