[MINC-users] Side-specific behaviour of subdivide_polygons

[Claude LEPAGE]

The two cases are different. When subdividing a "right" surface,
the polygons will be oriented in the opposite direction than left.

General rule: always re-orient the surface as a "left" surface 
before subdividing.

Also, the sphere that comes out from create_tetra is "left" oriented.

Here is my PERL wrapper for this:

# subdivide a surface taking into account if it's a left or right hemisphere.

sub subdivide_mesh {

  my $input = shift;
  my $npoly = shift;
  my $output = shift;

  my $npoly_input = `print_n_polygons $input`;
  chomp( $npoly_input );
  if( !CheckFlipOrientation( $input ) ) {
    &run( "subdivide_polygons", $input, $output, $npoly );
  } else {
    # flip right as left first before subdividing, then flip back.
    &run( "param2xfm", '-clobber', '-scales', -1, 1, 1,
          "${tmpdir}/flip.xfm" );
    my $input_flipped = "${tmpdir}/right_flipped.obj";
    &run( "transform_objects", $input,
          "${tmpdir}/flip.xfm", $input_flipped );
    &run( "subdivide_polygons", $input_flipped, $output, $npoly );
    &run( "transform_objects", $output,
          "${tmpdir}/flip.xfm", $output );  # flip.xfm is its own inverse
    unlink( $input_flipped );
    unlink( "${tmpdir}/flip.xfm" );
  }
}

# Check if the input surface has the same side orientation (left)
# as the default template model.

sub CheckFlipOrientation {

  my $obj = shift;

  my $npoly = `print_n_polygons $obj`;
  chomp( $npoly );

  my $ret = `tail -5 $obj`;
  my @verts = split( ' ', $ret );
  my @last3 = ( $verts[$#verts-2], $verts[$#verts-1], $verts[$#verts] );

  my $dummy_sphere = "${tmpdir}/dummy_sphere.obj";
  &run('create_tetra',$dummy_sphere,0,0,0,1,1,1,$npoly);
  $ret = `tail -5 $dummy_sphere`;
  unlink( $dummy_sphere );
  @verts = split( ' ', $ret );
  my @sphere3 = ( $verts[$#verts-2], $verts[$#verts-1], $verts[$#verts] );
  if( $last3[0] == $verts[$#verts-2] &&
      $last3[1] == $verts[$#verts-1] &&
      $last3[2] == $verts[$#verts-0] ) {
    return 0;
  } else {
    return 1;
  }
}


>
> Hi all,
>
> I was wondering if someone could explain if there are meaningful
> differences between the following two operations:
>
> $ subdivide_polygons   lowres_left.obj   highres_left.obj
> $ transform_objects   highres_left.obj   xflip.xfm   highres_right_0.obj
> vs
> $ transform_objects   lowres_left.obj   xflip.xfm   lowres_right.obj
> $ subdivide_polygons   lowres_right.obj   highres_right_1.obj
>
> Flipping then subdividing yields different results from subdividing then
> flipping:
>
> $ diff -q highres_right_0.obj highres_right_1.obj
> Files highres_right_0.obj and highres_right_1.obj differ
>
> I get the impression that the set of vertices is equivalent between the two
> hires surfaces, but there connectivity is somehow different.
>
> Thanks,
> Andrew
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>