насчет убогости я не согласен - по мне так вполне сносно, по сравнению с Облой так вообще замечательно.
Экспорт то есть. - вот на нексусе уже примеры
В его модельке вся иерархия правильная. Я думал что здесь знают может, или в Обливионе было чтото похожее.
Способ есть... но через одно место автогеном.
Лови:
[hide]
Not sure if this helps you guys... but noticed some values do peculiar things when importing OBJ's NiTriShapeData blocks (using latest Nif.xml in Nifskope)
Process:
Nifskope: Open a skinned character mesh .nif
Nifskope: Export as OBJ
3dsMax9: Import OBJ
3dsMax9: Export as OBJ
Nifskope (new instance): Import OBJ
pausing for a moment to observe the differences between the OBJ coming from Max's NiTriShapeData blocks, and the NiTriShapedata blocks in the original Nif
NiTriShapeData block from Max:
Num UV Sets 1 (original is 0)
BS Num UV Sets 0 (original is 1)
Has Normals yes (original is no)
Center and Radius slightly different (disregarding)
Vertex Color array is filled with mostly black, but alot of random verts colored bright white, green, yellow, pink (original is all black)
Consistency Flags CT_MUTABLE (original is CT_STATIC)
so, before copy and pasting the data blocks from one to the other, back in Max, i fill the Vertex Colors with solid black to every vert, the re-export the OBJ.
Now, The only difference at this point seems to be the exported obj has UV Sets, but no BS Num UV Sets, and has Normals.
Guessing the UV Sets vs BS Num UV sets is looking at UV channels 0-1 (id assume), and BS Num UV sets are the uv sets used for blend maps or such. Since OBJ format will only allow a single UV set to be exported, the others are stripped, so that seems to make sense there.
moving on:
Nifskope (new instance): Import OBJ (from max, now with vertex color array all black, offset Y(Z) height to -120
Nifskope (new instance): Match up as many of the values as possible before Copy n pasting
Nifskope (new instance): Copy NiTriShapeData block
Nifskope: Paste Over NiTriShapeData block
Observe value differences: (heres where it gets really messy)
Paste Over warning: Nif versions differ! Current File Version: 20.2.0.7, Clipboard Data version 20.0.0.5 The results will be unpredictable... (guessing this is why it goes wrong)
First Unknown Int goes to 1495937025 (was 0). Guessing this is the sum of the Normals Array values that erroneously got bumped up into that field
Normals array: X entries went to Z, Y entries went to X, Z entries went to Y
Center: X entry went to Radius, Y entry went to X, Z entry went to Y
Radius: entry went to Z on Center
Consistency Flags became 65535 (was CT_STATIC)
Additional Data became 474220908 (was None)
Num Triangles became 0 (was 2412)
Num Triangle Points became 16777217 (was 7236)
Has Triangles became No (was Yes)
Num Match Groups became 2 (was 0)
so, after I set the values where i could back to their original intended values, the mesh showed up properly.
Obviously the only things that were blown away now were the UVs apparently, which is why, when loading up that nif in game, the model shows up fine, but the textures are all skewed.
Anyways, guessing the values getting mucked up is just because of the version mismatch.
Just sharing my findings, not sure if helpful to anyone but, ehh.[/hide]
лично я - некуда не спешу и подожду нормального модуля для экспорта/импорта.
А если кому-то не втерпешь... можете экспериментировать.