|
|
Thanks for the comments everybody :)
I'm using a detail texture for the above-ground view, to avoid the blurry syndrom. I'm currently playing with the idea to actually use many detail textures, of different resolutions, to improve the quality (one detail texture is common, does anybody used more than one?)..
What you see on the ground is not a river, it's a road. I quickly said a word in the description about it: it is generated from a colored-base texture map which associates colors to materials. So, i basically have a heightmap (to get the height of the texels), and a colormap, to see if the texel must be procedurally generated (with the described algorithm), or taken from a standard texture. That way i can mix procedural and standard texturing in the same terrain, without additionnal polygons.
I haven't implemeneted LOD yet, btw, so can't answer for the XZ plane thing.
There is no shadows in the shot yet. Everything is per-texel lit. I'm playing with the idea to add per-vertex shadows, and since my engine already supports ray casting, it won't be a big deal to code in i guess.
Finally, my development machine is a simple P200 + Vaudoo 2. I get 15 fps for these 8000 tris, but i do not use triangle strips yet. With triangle strips the performance will probably double.
Btw, 150 fps is only 1.2 M triangles per second, and GeForces can reach much higher rates.
The terrain in the shot is 64x64 vertices, with a 512x512 texture. Yes, you can apply only some layers on specific parts of the terrain, using the color-based texture map technic i described above.
Source textures are tiled over the terrain, but with all the per-texel blending parameters, you don't actually see tiling (look at the shot and tell me how many textures were repeated? :))
|