Originally posted by lxnyce:
Wouldn’t this make a great personal competition.
Err, no.
Have you got nothing better to do, lxnyce?
Originally posted by lxnyce:
Wouldn’t this make a great personal competition.
Yeah, maybe I could make a homepage for this competition.
Originally posted by lxnyce:
If someone can provide a decent complex building (1 million plus polys) like hunsoul is proposing, we can all duke it out to see which algorithm is best to render it with. The model should be just a set of raw polygons, but complicated enough that we just can’t push the entire thing to the graphics card to render it (or stick in a display list). Preferably not just triangles, but in its original format. The model shouldn’t have any visibility information whatsoever or even textures for that matter (though it would be nice).
I could provide a very complex building - no problem. Maybe more than one, so we could test the algoritms not on one sample. The input file format could be - what I can provide - very simple. I can save the model in file format in which we a agree or .3ds file format (or maybe .ase or milkshape file format) but I think the .3ds file format is simple enough. I can save into the file only triangles or big polygons (convex, concav and holed polygons). Maybe I could put on the website of the models more than one version.
The competitors could push the advantage of the modern hardwares so the programs could be made for GeForce3 too but not for GeForceFX - only because I haven’t got GeForce4 so I couln’t test that programs on my machine. 
What do you think? I think it would be funny and very instuctive!
hmm i think such competition can be interresting, however, there are many things to think about.
-
is preprocessing data allowed?
-
how much max RAM may the program use?
-
how long can the preprocesing be?
-
LOD should be forbidden: too easy to get high framerates.
that’s all i have in mind at the moment…
1M triangle sounds quite little however…
Glad to see that other people don’t think this is a waste of time.
“1. is preprocessing data allowed?”
Sure. The objective is to see who can come up with the best generic algorithm for arbitrary building meshes. You can save the preprocess information, but I think that to be fare, everything should be run from scratch on any model given. There should be a general program skeleton to take in an input file and render it.
Kin dof like :
MyProgram.exe -input “Bulding1.txt”
“3. how long can the preprocesing be?”
As long as it needs to be, though I don’t think I or anyone else would sit 30 minutes or more waiting on a building to be preprocessed by your algorithm.
“2. how much max RAM may the program use?”
As much as you need to, but I think that will most likely be watched to see if the algorithm is usable or not for larger worlds (not just 1 but hundreds of buildings).
"4. LOD should be forbidden: too easy to get high framerates. "
If you can find a way to arbitrarily apply LOD onto a building of that size, I would say more props to you. Remember, the building is going to be just raw polygons or triangles. Each one of these is not guaranteed to have the same texture. All can possible use a different unique texture for all we know.
"that’s all i have in mind at the moment…
1M triangle sounds quite little however… "
If 1 Million is too little, then bump it up to 20 million. It all depends on the model we get. We just need a model we can’t shove as a display list or push to the hardware straight and get good frame rates. The model should also be tailored towards outdoor and indoor rendering, not like quake levels we often see. Think like a 100 story office building with offices on each floor, items in each room, etc…
As for the sources, I don’t think it should be mandatory to give it out unless the author wants to. However, I think that the algorithm should at least be gone over in full so that we can all benefit and judge for ourselves whether or not this method deserves any recognition.
I don’t want to give timelines or anything, because this is for fun. Since it is a competition (for bragging rights I guess), I feel that there will be some speed in seeing who can come up with something decent. So as to not sound like I am just trying to rip of someones algorithm, I plan on trying myself, even if my entry might suck (which I doubt).
To me this is fun. I love 3D/OpenGL, and visibility algorithms have always interested me. There is nothing to gain from this but experience and intellectual knowledge. If you don’t feel like doing it, then don’t, but please don’t knock the other people who like to do it for fun.