Well I guess my only option is to wait and watch for someone to try to do it. I do hope it will work. (Noone can say it will untill it has been done.) Is it 1.08 that will support it? or does 1.07 already support them?
Btw. (I know I could read it up somewhere :p) how many times bigger will the maps be? 4x in every dimension? 16 square, 64 cubic? Or actually 16x, 256 square, 4096 cubic? Well, it all comes back now, must be 16x, Krenzo said those would be aircraft only (since collision doesnt work prpperly below 1 unit [1 unit is critical already]).
There are also some examples in current maps, look at the 3dskybox part of MValley or Duststorm. That's 16x and both 3dskyboxes are made with the same level of detail than the main terrain. It wouldn't be possible to make a entire scaled map at that level of detail, but that's not needed anyway. A x16 map should use different scales of detail.
Thats great to hear. Though I do wish it were possible to do a full 16x map with all units. And not just aircraft. Were there any major problems you encountered while testing it?
Models aren't meant to be used like terrain if you mean a surface where units can move across. They could be used in mountains but the lack of proper shadows would add some new problems.
Mip maps are for textures. He's talking about making the terrain up of less triangles as you get farther away.
Ah. So what's the polygonal strain on a default computer (512MB, +2Ghz, decent mid-range graphics card like a Radeon 9800, etc.)? Actually, let me phrase it like this: How much more strain does a 16x map give (FPS-wise, server-wise) than the current ones, unoptimized?
What I would expect: Server - little to no difference Client - little to no difference It's all the same units, smaller models won't affect performance, especially since the map would be designed with the 16th part of the viewing distance of a 1:1 map.
It's all depending on the mapper. If the mapper isn't a fool then the map will run better or the same. And btw. .. the probability for LESS detail is higher, since the grid units are "longer" and such big maps will "want" farther viewing distance, forcing lower detail. People can build 1:1 Maps that run like shit and they can make 1:16 maps that run like shit, nothing new, nothing to care about. The mapper is responsible for map performance, 1:16 scale isn't.
Actually, if a standard map we have now is changed into a 16X map, it would be better. Why? Because the standard 1024 unit cuts will not be the same at the scale of 16, meaning a map the size of escort would be 16 times better, if optimized as well, i could estimate up to 30 times better.
16x maps should need much more work to look like 1:1, if mapper wants to make effective use of the whole area. That means bigger maps (bsp size), higher load times, higher memory requirements. If a map is going to have more displacements, brushes, models, etc, it also will need more optimization. Performance won't be good or bad because the scale, but a 16x map will tend to perform worse than a 1:1 map or it'll look completely bland. Also take in mind that some actual 1:1 maps aren't examples of optimization.
Idea: good use of large-scale decals on the terrain could help with some of the blandness while remaining relatively inexpensive. In other words, use a small handful of textures with minimal distinguishable parts (to reduce the appearance of tiling) scaled down 16 times as to match the scale of the player, then get some partially-transparent decals scaled way up and smattered across the terrain. It'll retain a high-res look from the ground, not look awful from way up in the sky, and not have draconian memory requirements. R_Yell, you're a skilled mapper, do you think that this idea would work?