This is a list of either open problems in computer graphics (or programming in general) that I find very interesting, or problems for which there exist good solutions but they involve so much (or so complicated) code that I don't have a solution of my own yet. Nonetheless I hope to one day have at least semi-functional implementations of all of them.

Volume/Surface Reconstruction From Point Clouds   Given a very large number of points that lie on the surface of a volume in 3-space, give me either a series of triangles that lie on this surface, or a series of tetrahedrons that compose the volume enclosed by these points.

Fluid/Solid Simulation   Given an arbitrary fluid and its velocity field, evolve it forward in time (and make it look like real water.) Or given an arbitrary volume represented as a series of tetrahedrons (say, a block of wood,) simulate it being deformed, cracked, fragmenting, and interacting with other solid volumes (or even other fluids.)

Raytracing   Simulate on an at least visually convincing level light transfer throughout a scene in reasonable (< 1 week) time. Hundreds of good solutions exist, but I have yet to implement any of them.

Joint Kinetics and Intelligent Balancing   First, physically simulate a series of joints and "muscles" connected to a "body." For example, a human mesh with a bone structure underlying it should fall down if its muscles aren't exerting the right forces to keep it upright, or if its limbs aren't in the correct position to balance the human. After that's done, create an AI that can intelligently get from place to place or position to position without losing its balance, under the constraints of this physical simulation, by controlling only the forces generated by the body.

last updated: 09 May 2005   © 2005 by Matthew Fisher