SharkD nous apporta ses lumieres en ce 2008/02/08 21:45:
> Alain <ele### [at] netscapenet> wrote:
>> Typicaly, the Sun is at a very large distance from your scene objects. It have
>> an aparant radius.
>> You use the "parallel" keyword and point it to the center of your scene. This
>> simulate a light source that is very far away.
>>
>> You use an area_light with circular. You can also use "orient" to make sure the
>> light aray is always perpendicular to the location tested.
>> circular with orient create a "spherical" area_light.
>>
>> If you can see your sun, directly or trough reflection(s), add a looks_like to
>> make it visible.
>>
>> --
>> Alain
>> -------------------------------------------------
>> To define recursion, we must first define recursion.
> 
> The problem with this method is that objects that don't lie along this vector
> are not lit properly. I would prefer a method that lights objects properly
> regardless of their position relative to the sun.
> 
> 
Unless your scene cover an extremely whide area, parallel is not a problem.
If your scene extend in an area large enough for that to be a problem, then you 
will have more problems with floating point precision: your objects will become 
to small relative to the distances separating them.
In real life, and using a scale equivalent to the earth-moon distance, the 
direction difference for the sun will be effectively negligeable, at around a 
degree of angle. The earth travel less than a degree per day, and that's much 
more than it's diameter, or the orbital diameter of any artificial satellites.

If your scale is such that you have several planets showing, then, using any 
kind of area_light is overkill exept for close objects in the foreground, as the 
bigest planets, like Jupiter and Saturn, will only show as single pixel, or 
sub-pixels dots.

-- 
Alain
-------------------------------------------------
Never underestimate the power of stupid people in large groups.

Post a reply to this message