3D Representations

Tags

computer-graphics

There are many approaches:

• Polygons
• parametric patches
• CSG
• space subdivision
• implicit representation

See [see page 28, here] for a list of considerations to take into account when choosing one representation over another.

Designing 3D Models

There's [see page 22, numerous] ways to design 3D models:

• 3D Digitiser - create a model IRL and then scan it into a digital representation.
• Laser Ranger - Like 3D Digitiser except instead of manually going over the shape of the model, you use a depth sensing camera to automatically pick up the shape.
• Photogrammetry - Take pictures of an object from a bunch of different angles and then use that to generate a mesh for the object.
• Modelling Software - Edit mesh directly (blender) or sculpt objects (zbrush).
• Mathematical Description - Use sin/cos waves to create a wave like structure.
• [see page 27, Sweeping] - Given a curve representing how a point varies across the surface of an object, you can fill in that point as you sweep through an axis. Consder a potters weel where we spin the clay and then shape both ends of it to make an object. This approach has [see page 30, issues] with:
  • Arch Length Parameterisation
  • Curvature
  • Reference Frames (eg. Frenet Frames)
• Procedural Techniques - Parametrize surface attributes to create variations on a model. You get one big shape, generate a bunch of smaller shapes ontop of it (and then on top of those shapes, recursively) until you get a final object. See [see page 33, Fractals].

We can optimize rendering of objects using see page 34, level of detail. I.E. we lower the number of polygons relative to the distance from the object. See progressive meshes.