Yes, unfortunately I can't do much about that, t3d for old Unreal/UT99 doesn't export package.group.name but only name; the only workaround is to export what the map actually needs, but this would require extra effort.___ wrote: ↑Mon Jan 27, 2020 6:20 pmI managed to export all the textures with UModel, but noticed that subdirectories in t3d_tex aren't searched during the conversion. So basically it seems to assume a flat hierarchy of files, but this is a bit of a problem since package X and package Y can both have a texture with the same name which causes a conflict. The .t3d file actually doesn't seem to specify the texture package either so maybe this is a limitation of the format?
Right now you have to merge-copy all subfolders into one to make it work as you probably figured out yourself.
I should probably scan the texture folder for all the textures recursively, but I'm afraid that it would slow down the importer (imagine thousands of files in various subfolders)
I've noticed, however, that UT2004 exports t3d textures as package.group.name; but the problem is that most of the detail world geometry is composed of static meshes, which are not saved in t3d
I think UT2004 can export to obj directly, but the UVs are messed up as far as I can remember and it exports geometry for all items as well, so you get a huge mess with a ton of triangles.
UModel doesn't export animated textures (this is why you get purple water) and exporting animated textures in UnrealEd crashes the editor.
However, it's possible to look up the base texture and simply copy that as the animated one.
As for UVs: if you provide the importer with textures, then I'll parse image headers to determine texture dims, this is why it's not messed up.
Other tools typically require to either transform permanently (this already messes up the UVs) or make a huge brush and intersect with the whole map,
but this creates lots of holes and breaks CSG in my experience, probably due to semisolids. Turning semisolids into solids may help, but I remember
having CSG robustness issues in this case as well (plus you get way more polygons).
So thanks again for the feedback, I'm happy to see that it works for you.