> If it looks right in one limited test case, doesn't mean it's right, or > the proper way to deal with it (so it won't backfire later on)
Look, basic ray optics says that the error in ray propagation I make by neglecting the curvature is linear proportional to the error in normal angle which I make by neglecting curvature. So in order to estimate the error in reflection, I need to estimate the error in normal angle. If I go a distance d, the normal angle changes by arctan (d/R) where R is the radius of earth. For any rendering distance we can achieve d/R is a very small number O(100km / 6000km) ~ 0.016. For small arguments, I can expand arctan(x) ~ x, so the angle changes by 0.016 or 0.9 degrees at the edge of the visibility. In the region where we actually see details of wave patterns and reflection, the error is a factor 10 less. However, on top of the normal I add wave noise, and the distortion of the normal by a wave can easily be 30-45 degrees. Compare this with 0.1 degrees error in the near range due to neglected curvature and you will come to the conclusion that any halfway realistic wave pattern easily masks any curvature errors I could possibly make with really high accuracy. This is mathematics. Not a limited test case. Plus, if you neglect the curvature effect in every relevant vector, the rendering artefacts at the tile boundaries must go away, because the differences in rendering geometry between tiles go away, and they're the only thing which can introduce the artefacts in the first place. There's simply no point in trying to solve a problem exactly if you can show by error analysis that a Taylor approximation gets you the desired result much faster and cheaper. Trust an old physicist on that point... :-) * Thorsten ------------------------------------------------------------------------------ For Developers, A Lot Can Happen In A Second. Boundary is the first to Know...and Tell You. Monitor Your Applications in Ultra-Fine Resolution. Try it FREE! http://p.sf.net/sfu/Boundary-d2dvs2 _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
