SharkD schrieb:
> Yes, but precision errors occur if I make the spheres *too* big. How big 
> would be big enough? In my scene 1 unit = 1 meter.

Don't worry about precision issues.

For exact scale, the atmosphere should have a radius of some 6e+6 units, 
which is indeed comparatively close to the distance limit at which 
objects disappear; still, it's within that limit. Also note that this 
distance cut-off is /relative/ to a ray's origin (e.g. the camera), and 
with an atmosphere thickness of just a few 1000 units and the camera 
just a few units away from the ground, you'll only be able to see a 
small section of the atmosphere anyway. Somewhere between 1e+5 and 1e+6 
units I'd guesstimate - dunno.

Bounding is no issue either: You'd be inside the sphere anyway, so the 
bounding box will be intersected anyway, so even if it is of low 
precision or turned off entirely, that doesn't make a difference.

And don't worry either about possible loss of details due to the center 
of the atmosphere being quite far away from your details: As POV-Ray 
uses double-precision math, a scene encompassing the whole earth would 
still allow headroom for details as small as a nanometer(!), so there 
should be enough headroom.


Make sure though to center your area of interest at <0,0,0>, so don't 
make that the center of your atmosphere.

>> Another thing that looks weird is the apparent "stacking" - hard to 
>> tell where that comes from. Maybe the geometry is way different from 
>> what I think it is.
> 
> Not sure what you mean by "stacking".

Well, the clouds look somewhat like the closer(?) ones are thinned-out 
copies of the further-off(?) ones. Or as if they were actually as far 
off as what they appear to be copying, just higher.

Post a reply to this message