1,360
edits
Line 86: | Line 86: | ||
Compared to almost anything else, the performance impact of re-positioning the sphere every frame is minimal. | Compared to almost anything else, the performance impact of re-positioning the sphere every frame is minimal. | ||
The reason is that graphics cards operate with floating point precision, and a sphere with the real size of Earth would create (potentially architecture dependent) massive numerical problems (the default far plane clipping distance in FG is some 120.000 m). The same is in fact true for the skydome - it's actually a half dome some 20 km away from the camera, made to appear behind everything else by rendering magic. Or, in other | The reason is that graphics cards operate with floating point precision, and a sphere with the real size of Earth would create (potentially architecture dependent) massive numerical problems (the default far plane clipping distance in FG is some 120.000 m). The same is in fact true for the skydome - it's actually a half dome some 20 km away from the camera, made to appear behind everything else by rendering magic. Or, in other words, we can't simply render a 'real earth' and a 'real atmosphere' because that would potentially overtax the GPU - real time rendering is always trickery of some sort. | ||
The implementation has the negative consequence that you may not move the view point too far from the spacecraft (the ray optics illusion works for one object, but not for two simultaneously) - in particular flyby view or tower view will give you odd results with Earthview. | The implementation has the negative consequence that you may not move the view point too far from the spacecraft (the ray optics illusion works for one object, but not for two simultaneously) - in particular flyby view or tower view will give you odd results with Earthview. |
edits