many methods of PShape do not work with loaded obj files

[sterlingcrispin]

IMHO all of the listed methods of PShape should work with loaded OBJ files

using processing 2.0.3 and 2.1.1

this code is modified from LoadDisplayOBJ example, many of the methods listed here do not work with this loaded OBJ file http://processing.org/reference/PShape.html , as an example below, getVertexCount() returns a zero

PShape rocket;

public void setup() {
size(640, 360, P3D);
rocket = loadShape("rocket.obj");
println(rocket.getVertexCount());

}

[hamoid]

Did you explore the object with getChildCount() and getChild() to make sure there's no vertices contained in the children? Which ones are exactly "many of the methods"?

[sterlingcrispin]

As I said, getVertexCount doesn't work, I didn't test every method once they started failing.

This is a case of the documentation being unclear, I posted this in the documentation area and was told it was a bug which should be posted here.Nothing in this reference describes the functionality of loadShape cutting an OBJ file into groups by individual faces and storing them in an array which must be accessed via getChild(). I was operating under the assumption that you could access an array of vertices that contained every vertex of the OBJ file rather than on a per-face basis, which is counter intuitive and unlike any other library I've ever used.

http://processing.org/reference/PShape.html

the following code works to get every vertex of a loaded OBJ file:

PShape rocket;

public void setup() {
size(640, 360, P3D);
rocket = loadShape("rocket.obj");

for(int i = 0; i < rocket.getChildCount(); i++){
for(int j = 0; j < rocket.getChild(i).getVertexCount(); j ++){
println(rocket.getChild(i).getVertex(j));
}
}

}

the following code is how I would have assumed this class worked based on the reference:

PShape rocket;

public void setup() {
size(640, 360, P3D);
rocket = loadShape("rocket.obj");

for(int i = 0; i < rocket.getVertexCount(); i++){
println(rocket.getVertex(i));
}

}

[hamoid]

Does that happen with every .obj? Or is it just that one?

[sterlingcrispin]

perhaps you can load the example in question and open a few OBJ's you have, it appears to be with every OBJ

[hamoid]

I tried with a few objects and yes, apparently each face is placed in
it's own child. I think that means that getVertexCount works correctly,
just not as you expected, or? It could count vertices recursively, but
that's another topic. As it is, the root object contains no vertices,
and the children contain 3 each.

Maybe the object loader could be changed to avoid creating all the extra
children, and have the vertices in the root object instead. But there
may be a reason for the current implementation.

Someone with more experience should answer that.

[codeanticode]

Yes, the OBJ parser creates a separate child shape for each face, the main reason begin to allow faces to have separate materials (because Processing only supports single texturing of shapes), and also because each face is not necessarily of the same kind (triangle, quad, general poly). The code could certainly be optimized to combine faces when they are of the same kind or they have the same material. Pull requests are welcome :-)

Anyways, what other methods don't work as expected?

[sterlingcrispin]

Thats a great explanation of this issue and should be included in the
documentation somewhere

On Mon, Nov 3, 2014 at 11:48 AM, codeanticode notifications@github.com
wrote:

Yes, the OBJ parser creates a separate child shape for each face, the main
reason begin to allow faces to have separate materials (because Processing
only supports single texturing of shapes), and also because each face is
not necessarily of the same kind (triangle, quad, general poly). The code
https://github.com/processing/processing/blob/master/core/src/processing/core/PShapeOBJ.java
could certainly be optimized to combine faces when they are of the same
kind or they have the same material. Pull requests are welcome :-)

Anyways, what other methods don't work as expected?

—
Reply to this email directly or view it on GitHub
#2873 (comment)
.

http://www.sterlingcrispin.com

[codeanticode]

Yes, you are right.

@alignedleft We should add a brief note to the docs explaining the quirks of the OBJ parser. Should I simply edit the PShape entry, or is this too specific?

[hamoid]

Wouldn't https://processing.org/reference/loadShape_.html be a better place? I think that's where you may encounter the issue.

[shiffman]

I looked at this and here is another technique that works. I agree this needs to be resolved in the documentation. I'll add an issue there.

  PShape obj = loadShape("test.obj");
  PShape tes = obj.getTessellation();
  int total = tes.getVertexCount();
  for (int j = 0; j < total; j++) {
    PVector v = tes.getVertex(j);
    point(v.x, v.y, v.z);
  }

[shiffman]

@benfry is adding a warning for this now (if you call getVertexCount() on a GROUP of PRIMITIVE) and I am on the documentation.