I wouldn't bet that someone will finish it. It's not too hard to trick around with the script, the line in question is while cnt < 5 : For l_bell_d.xbx you can replace 5 by 21, but some sub meshes come out ugly then because they require noesis.RPGEO_TRIANGLE_STRIP (So there must be a flag in the mod...

For the xmls I assume once you know which object (from actorentry.xml) has
object UID="32651"
you can use this as position offsets: (just a wild guess)
<element index="12" data="18000.000000" />
<element index="13" data="0.000000" />
<element index="14" data="6300.000000" />

Don't have those textures but might stand for "diffuse map" (or dark mage, whatever ).

Oh, shxt, can't remember having ever seen a 3D format using double for the vertices...
.

MinXWin wrote: ↑

Tue Dec 15, 2020 11:16 pm

I wonder if the duplicates could be the UV coordinates been some how packaged along side the verts...

Guess 'no' (near to 100% ), search for array/tex_uv_0, there's the uv data to be found.

No, I didn't use MR - some C coding on the fly using floats. But obviously I'm missing something - the vertices need a special treatment, I guess. Maybe sort them by those "indices", 0x20, 0x40, 0x80...? Or use a threshold (that worked for a 3D format with a similar problem), at no avail here. Also ...

Guess, you'll need to use Bigchillghost's AMR. . 116_hex.png This is for "hanging platform", black rectangles contain the first vertex (doubles). edit : had to correct this from "floats" to "doubles" (big thanx to Bigchillghost for the hint!) Face indices start at 0x31B9 (DWords, 4 bytes). Count is ...

4000-4500

Nice script! Position offsets would be nice, too, to not have the meshes clumped:
.

Using hex2obj (view link in my sig)
. (One submesh only.)

Getting some vertices (search for "array/vertex") from 116 file ("hanging platform"): . hanging_platform.png Seems there's an issue with the faces or - more likely- I accidentally used vertex colors for scaling (per vertex!, hahaha). Without this it looks different (right part of pic) but still not ...

Several smaller submeshes, no time for fiddling.

Here you go: . birdhouse_point_cloud.png Hex2obj will give you a point cloud only. Getting the face indices requires additional coding which I did "on the fly": # f 1 5 4 f 2 3 6 f 2 6 5 f 3 2 1 f 6 4 5 f 7 8 12 f 12 11 7 f 8 10 14 f 14 12 8 f 10 9 13 f 13 14 10 f 9 7 11 f 11 13 9 f 15 17 18 f 18 16...