That was indeed the problem -- the level of detail was miscalculated
because it was based on "radius", which an ellipsoid doesn't have.
I've updated the github repository with new versions of
sphere.hpp/.cpp and ellipsoid.hpp/.cpp, which fix the problem.
I haven't yet built and posted new installers, which will be VPython 5.74.
Bruce Sherwood
On Sun, Apr 22, 2012 at 5:07 PM, Bruce Sherwood <Bru...@nc...> wrote:
> It has nothing to do with the size of the ellipsoid. VPython handles
> numbers with an enormous range; 3400 isn't a big number. The ellipsoid
> object inherits from the sphere object, and you'll see that if you
> change your ellipsoid to a sphere it will look fine.
>
> I was not previously aware of this problem with ellipsoids. Probably
> the bug is in src/core/sphere.cpp, in the gl_render routine. It must
> be that the "level of detail" (lod) is being calculated incorrectly,
> as though the ellipsoid occupied a very small portion of the window
> (in which case a sphere or ellipsoid is rendered crudely, since then
> you can't see detail). Maybe the mistake is that the screen coverage
> is computed using "radius", which is an attribute of a sphere but not
> of an ellipsoid.
>
> Bruce Sherwood
>
> On Sun, Apr 22, 2012 at 4:11 PM, K.-Michael Aye <kmi...@gm...> wrote:
>> Dear all,
>>
>> I noticed a lack of precision in the rendering of the ellipsoid when going
>> to big numbers.
>> I need these big numbers because I am trying to display planetary remote
>> sensing illumination scenarios.
>> Here is the code I used to create the attached image:
>>
>> from visual import *
>>
>> xaxis = arrow(axis=(4000,0,0), shaftwidth=40, color=color.red)
>> yaxis = arrow(axis=(0,4000,0), shaftwidth=40, color=color.green)
>> zaxis = arrow(axis=(0,0,4000), shaftwidth=40, color=color.blue)
>> mars = ellipsoid(length=2*3396.19,height=2*3396.19,width=2*3376.2,
>> opacity=0.5, color=(1,0.7,0.2))
>>
>>
>> As you can see the ellipsoid is very 'edgy', can this be improved somehow?
>>
>> Best regards,
>> Michael
|
