Some advice on Triangle.NET.

Nov 7, 2016 at 7:51 AM
Good library! And I have some advice here.
  1. Make Triangle.NET Objects/APIs suitable for GPU rendering instead of the CPU methods in the demos/examples.
  2. Add a obj exporter. (Obj is a text-based 3D model file format.)
  3. Move the project to GitHub! It's easy!.
  4. Keep the focus on the core functions. The demo is cool but I think there are more things left to do with the core Objects (Vertex, Polygon, ...) and APIs.
    For example, class shouldn't be used for small object. I should say class is abused in Triangle.NET. More infomation on that here.
  5. What's the advantage compared to poly2tri-cs? Show the tests and profile statics in a documentation page please.
Nov 8, 2016 at 10:58 PM
  1. I think it would be fairly difficult to rewrite the classes to match vertex or index buffers for GPU. That should be kept separate (the TriangleNet.Rendering project already contains some buffer helpers, which I use with OpenGL).
  2. Nope. There are just so many different file formats. And Triangle isn't meant for 3D anyway.
  3. Not at the moment (though I do prefer GitHub, too).
  4. I'm aware of the issue (see, but that would probably mean rewriting major parts of the library, which I can't do right now.
  5. Well, the differences should be pretty obvious. Never did any benchmarks, though.