Am 08.01.2016 um 00:46 schrieb Mike Horvath:

> So, area lights are cheaper than an equally sized array of point lights?

Yes, absolutely: With an array of point lights, POV-Ray needs to perform
shadow tests against every one of them every time. With an area light,
POV-Ray can choose to ignore some of them if it finds that their
neighbors are all shadowed or all unshadowed. It's kind of like
anti-aliasing.

You'll need to use the "adaptive N" keyword to enable this optimization,
where N governs how many of the sub-lights POV-Ray will always test:
POV-Ray will initially render only (2^N)+1 of the sub-lights in each
direction, then render the sub-lights halfway between those that gave
noticeably different results, and then in the same way "drill down" ever
further until it reaches the level of maximum resolution you specified
for the area_light.

Another way in which area lights can speed up rendering is that normally
they only affect shadow testing, while still behaving as a single point
light for diffuse and highlight computations. (If that is undesired, for
instance because the light source is very close to a surface, you can
use the "area_illumination on" keyword to make the area light affect
diffuse and highlights as well.)

And the third way in which you can benefit from area lights is the
"jitter" feature, which can reduce banding artefacts in soft shadows
without having to increase the number of sub-lights.

Post a reply to this message