Max poly count

#1 Um, sure. If you don't run out of memory first your framerate will probably be measured in frames/hour, though.

#2 This is what LOD is for, I suppose. You'll have to go with static LOD, because for 2000 models it's going to be too expensive (CPU-cycle-wise) to make use of CLOD.

Texture storage is going to present a problem as well. A lot of state switching isn't going to be a pretty thing either.

#3 >>Texture storage is going to present a problem as well

even if you reuse textures? thinking in terms of bots here, not pcs each with an individual skins. maybe 20-40 models, 40-80 skins, only eight or so actual player characters. using flocking for the ai, and extensive lod (is static load pre-fixed lod levels?), would it be possible?

>>A lot of state switching isn't going to be a pretty thing either.

you mean the switches for all those models? would flocking/coding ai as units help here?


i need to know how to make crowd scenes possible. i figure most of the bots will follow simple scripted behavior until they interact with a player. since the number of bots able to interact with a player at any given moment will be relatively small, they shouldnt cost too much. i guess im looking at some advanced lod programming, eh? would extensive use of zone portals help too?

what i really need are numbers for polys in t2/Torque. how many for the maps and how many for the models. im sure i can find more about lod elsewhere.

#4 i was also thinking of scripting much of the crowd's behavior so that nothing needs to go out over the net except for what directly interacts with the players

#5 the texture manager keeps a library of textures that are used, so it won't have to precache 1 texture several times. Therefore I believe using the same texture for many models will speed your scene(and especially loading time of the map) up a little.