Talk:Catmull–Clark subdivision surface

What's the input for the function? Example input/output?

typical 3D datas are formated like this: an array of vertex points first (x,y,z) 3D coordinates with type (float, float, float), they are indexed from 0 to (n - 1), then a list (or array) of faces, a triangle face is specified with 3 ints, and a quad face with 4 ints, the ints are the indexes in the first array. This data model is most often used because vertices belongs to several faces, so there are no duplication of the coordinates with this model.

here are below a typical input and output for the Catmull-Clark subdivision surface algorythm, the input is a simple cube:

input_points = [
  (-1.0,  1.0,  1.0);
  (-1.0, -1.0,  1.0);
  ( 1.0, -1.0,  1.0);
  ( 1.0,  1.0,  1.0);
  ( 1.0, -1.0, -1.0);
  ( 1.0,  1.0, -1.0);
  (-1.0, -1.0, -1.0);
  (-1.0,  1.0, -1.0);
]

input_faces = [
  (0, 1, 2, 3);
  (3, 2, 4, 5);
  (5, 4, 6, 7);
  (7, 0, 3, 5);
  (7, 6, 1, 0);
  (6, 1, 2, 4);
]

# below after one iteration:

output_points = [
  (-0.555556,  0.555556,  0.555556);
  (-0.555556, -0.555556,  0.555556);
  ( 0.555556, -0.555556,  0.555556);
  ( 0.555556,  0.555556,  0.555556);
  ( 0.555556, -0.555556, -0.555556);
  ( 0.555556,  0.555556, -0.555556);
  (-0.555556, -0.555556, -0.555556);
  (-0.555556,  0.555556, -0.555556);
  ( 0.000000,  0.000000,  1.000000);
  (-0.666667,  0.000000,  0.666667);
  ( 0.000000, -0.666667,  0.666667);
  ( 0.666667,  0.000000,  0.666667);
  ( 0.000000,  0.666667,  0.666667);
  ( 1.000000,  0.000000,  0.000000);
  ( 0.666667, -0.666667,  0.000000);
  ( 0.666667,  0.000000, -0.666667);
  ( 0.666667,  0.666667,  0.000000);
  ( 0.000000,  0.000000, -1.000000);
  ( 0.000000, -0.666667, -0.666667);
  (-0.666667,  0.000000, -0.666667);
  ( 0.000000,  0.666667, -0.666667);
  ( 0.000000,  1.000000,  0.000000);
  (-0.666667,  0.666667,  0.000000);
  (-1.000000,  0.000000,  0.000000);
  (-0.666667, -0.666667,  0.000000);
  ( 0.000000, -1.000000,  0.000000);
]

output_faces = [
  ( 0,  9,  8, 12);
  ( 1, 10,  8,  9);
  ( 2, 11,  8, 10);
  ( 3, 12,  8, 11);
  ( 3, 11, 13, 16);
  ( 2, 14, 13, 11);
  ( 4, 15, 13, 14);
  ( 5, 16, 13, 15);
  ( 5, 15, 17, 20);
  ( 4, 18, 17, 15);
  ( 6, 19, 17, 18);
  ( 7, 20, 17, 19);
  ( 7, 22, 21, 20);
  ( 0, 12, 21, 22);
  ( 3, 16, 21, 12);
  ( 5, 20, 21, 16);
  ( 7, 19, 23, 22);
  ( 6, 24, 23, 19);
  ( 1,  9, 23, 24);
  ( 0, 22, 23,  9);
  ( 6, 24, 25, 18);
  ( 1, 10, 25, 24);
  ( 2, 14, 25, 10);
  ( 4, 18, 25, 14);
]

I propose this image to illustrate the article:

Looks good, but I would suggest breaking it into three images and losing the built-in descriptor lines. I would also suggest stretching the contrast ratio such that it's white on black, rather than light-gray on dark-gray. --Michael Mol 08:08, 9 January 2010 (UTC)

Algorithm Description Improvements

I've tried to improve the description of the algorithm (making it look more like math and less like computer code) but may have made a few bloopers along the way. A full mathematical version would be even better, except that I doubt that most people would immediately pick up the notation for a centroid and so would go horribly astray. Still, if someone can do some more work on it I'd be glad. Note also that neither the Wikipedia article nor this task are particularly good references; for example, the WP article only handles the hole-free mesh case. –Donal Fellows 15:19, 18 January 2010 (UTC)