The previous post on this subject, New discussion on UD (updated) , has lots of interesting comments. It has become difficult to make sense of the various proposals concerning the UD algorithm, therefore I made this new post which shall serve first as a place for new discussions, then it will be updated.
It is maybe the time to make sense a bit of the material existent (or linked to) on this blog. That is why I invite everybody who is willing to do this to make it’s own proposal, preferably with (either) programs or proofs or evidence (links). Especially, in a random order, this invitation is addressed to:
- Dave H
but anybody which has something coherent to communicate is welcome to make a proposal which will blow our minds.
Make your respective proposals as detailed as you wish, take your time to make them convincing and then, if you agree, of course, we shall make feature posts here with each of them, in order to become easier to follow the different threads.
Let me finish, for the moment, by stating which points are the most important in my opinion, until now (I stress, in my opinion, feel free to contradict me):
- UD works like a sorting algorithm,
- cubemaps are used, but their main utility is to eliminate rotations wrt the POV from the problem,
- UD is not a rendering algorithm (or, at least, the most interesting part of UD is not one), it is an algorithm for fast searching the data needed to put a colour on each pixel,
- UD needs to turn a cloud of points into a database, only once, operation which takes considerable more time than the searching algorithm part,
- does not need antialiasing
- does not raycasting for each pixel
- it is a mathematical breakthrough, though not a CS or math technical one (i.e. does not need the latest edge CS or math research in order to make sense of it, but it’s a beautiful idea).
Almost finished, but I have to explain a bit my attitude about UD. I am thorn between my curiosity about this, explained in other posts (for example by it’s partial overlap with the goals of Digital Materialization), and the fact that I don’t want this blog to be absorbed into this subject only. I have my ideas concerning a possible UD algorithm, especially from a math viewpoint, but unless I produce a program, I can’t know if I’m right or wrong (and I am not willing to try yet, because I am sympathetic with the underdog Dell and also because it would take me at least several months to get off the rust from my programming brain and I am not yet willing to spent real research time on this). Suppose I’m wrong, therefore, and let’s try to find it in a collaborative, open way. If we succeed then I would be very happy, in particular because it would get out of my mind.