A bunch of comments:
First and foremost. Thank you very much, Carsmaster, for continuing to push the limits!
Second. Thanks, Despereaux, for jumping in with alternate versions to help those having lesser computers. You beat me to it!
Third. My approach in utilizing varying transparency by LOD is to make for something of a transition in appearance. For two reasons:
1) So that at the farthest visible distance the grass doesn't suddenly pop in or out of visibility like a switch being thrown.
2) So that when the grass is placed against differing landscape or runway textures it doesn't stand out too strongly at farther view distance.
By the foregoing, I'm saying that the grass structured as first presented is basically fine when used against a particular set of terrain and runway textures. When tuned for appearance against one particular, underlying texture set, it can be jarring when used on some other rather different underlying textures, in one or both of tone and hue.
Fourth. (And this will be a long one.) I played with the lighting parameters in the .mat files (Ambient and Diffuse) so as to find some reasonable consistency in general appearance against the terrain/runway as the Sun elevation changes. It can't be perfect because of the combination of the textures being flat planes and that the surface normals aren't optimized for this fact of the flat-plane simplifying of complex objects.
Take the flowers, for example. Let's for simplicity assume all their textures are exactly vertical (they're close enough to that right now), being a flat sheet standing on edge. The surface normals for the 4 corners are thus oriented horizontally, perpendicular to the face. This is the case for both sides, with the opposing normals pointing in opposite directions.
Let's follow what happens to the object lighting as the Sun varies in elevation, for a texture facing east. The lighting will be strongest when the Sun is on the eastern horizon, at sunrise. That's because the surface normals are pointing directly at the Sun. As the Sun climbs, the lighting decreases immediately, and continues to a functional minimum at noon, when the Sun is in plane with the east-facing side. From there until sunset the lighting remains minimal.
Now here's the thing. A real plant is not a flat plane. It has surfaces in a variety of orientations. And so if we look at one side all day we never find that the plant will be darkest when the Sun is highest in the sky! Indeed, to a general approximation we find a not large change in overall illumination throughout the whole day.
How do we work around that here? By NOT making the surface normals perpendicular to the polygon faces!
For example. Here's both facese from a 2-sided, 4-corner, 15m x 15m forest face I have in my objects:
0 -15 0 0.342 0 0.9397
0 15 0 0.342 0 0.9397
0 15 15 0.342 0 0.9397
0 -15 15 0.342 0 0.9397
0 -15 0 -0.342 0 0.9397
0 15 0 -0.342 0 0.9397
0 15 15 -0.342 0 0.9397
0 -15 15 -0.342 0 0.9397
A naive (and incorrect) approach with surface normals in the usual perpendicular orientation would be:
0 -15 0 1 0 0
0 15 0 1 0 0
0 15 15 1 0 0
0 -15 15 1 0 0
0 -15 0 -1 0 0
0 15 0 -1 0 0
0 15 15 -1 0 0
0 -15 15 -1 0 0
What I'm doing here is radiacally altering the normals so that instead of pointing horizontally, they're pointing almost straight up, just 20 degrees from vertical. This ensures that lighting of this forest object will be generally highest when the Sun is highest, but with still some emphasis on that side facing the Sun. This works beautifully, considering the severe limitations of the object simplicity.
I'll look at this situation for the flowers, and see if I can easily apply such alterations via a spreadsheet. (I don't have any proper 3D software.) The grass doesn't need this treatment because the polys are already tilted a lot, thus making for reasonably good lighting characteristics.
Author
Topic: RELEASE CARSMASTER GRASS v.1.0 (Read 23033 times)
