What steps will reproduce the problem?
1. Download the 3DS file http://www.baument.com/restore/Models/Faucet1.zip
available at http://www.baument.com/restore/archives.html
2. Read it with lib3ds
3. In smooth normals computation part of lib3ds_mesh_calculate_vertex_normals
function of lib3ds_mesh.c, this should lead to a few incorrect normals (I say
"should" because I got this result using a pseudo-translation of lib3ds in Java
3D).
What is the expected output? What do you see instead?
Getting default normals from lib3ds_vector_normalize when length is 0 isn't a
major issue, but it would be better to keep the default normal in that case
(i.e. the normal computed before smoothing operation). After line 222 of
lib3ds_mesh.c, I would suggest to add this test:
if (lib3ds_vector_length(n) == 0) {
lib3ds_vector_copy(n, fa[3*i+j].normal);
}
(with an additional lib3ds_vector_length_squared function, you could spare a
call to sqrt here)
What version of the product are you using?
I looked to source code of version 20080909
Original issue reported on code.google.com by epuyba...@gmail.com on 16 Sep 2014 at 9:42
Original issue reported on code.google.com by
epuyba...@gmail.comon 16 Sep 2014 at 9:42