Greg Ward wrote:
Hi Barbara,
You could use a smaller polygon for a light shelf, but there is less benefit from doing so. You get a more efficient calculation in the case of blinds simply because you don't have to compute and place as many ambient values if you restrict mkillum to a smaller area, yet the result (provided there is no nearby exterior geometry) will be just the same.
Users may find modelling blinds with direct sunlight benefits from the photon-map add-on, since forward raytracing is more effective than backward raytracing in this situation. At least when using the entry aperture option in photon-map. The photon map method avoids problems with mkillum's glowing area surfaces as well.
-Peter
PS: The basic idea is to trace only lightpaths which carry most of the scene's energy. A low energy ray is a wasted ray in terms of computer time. That's why modeling blinds with ambient rays only on a sunny day is such a drag: Most rays miss the window, those traced through the structure miss the sun and hit something low lit outside (or inside, depending on the blind). On a cloudy day it actually doesn't work as bad.
Mkillum intruduces an auxilary surface and does the same backward step from this surface. This speed things up, but doesn't work well for glazings of complex structure (multiple reflecting elements in it), is limited in angular output resolution and is based on self luminous area sources (which are subdivided). The latter two can be eased with parameter setting, but the concept is not nice for elements generating a highly directed output (light redirecting elements without diffuser).
Photon-map was motivated to cope with these: Tracing rays from the sun (actually from the bounding cube, see Roland's phd for details) though the window structure (including mulitple reflecting surfaces) into the room. It solves the problem of interfacing the forward with the backward step more elegantly, as the traced photons are stored on the inside room surfaces and not at the window. If you're prepared to face some new parameters for photon-map, you'll find it worth the effort.
If anyone locates money and workers, there's another interesting step beyond that: more effective raytracing based on the Metropolis algorithm.
- end of compact early-morning Radiance comment -
···
--
pab-opto, Freiburg, Germany, www.pab-opto.de