C. Cappai <nomail@nomail> wrote:
> Full area light!

  Perhaps "full" was not a wise choice of word, as it can be slightly
misleading. My original intention was for it to mean "everything
related to rendering the effects of light sources, that is, shadows and
diffuse/specular lighting, can now take into account the area light
parameters". (Previously only shadows took into account area light
settings.)

  I understand now that someone could perhaps mistakenly think that
"full" means area lights are now simulated as true area lights, ie.
lights with a non-zero surface area, instead of an (optimized) grid
of point lights.

  That's of course not the case. Area lights are still a grid of point
lights, which are (smartly) sampled. The difference is that the sampling
can now be turned on for diffuse/specular lighting too. This means that
the Size_1 and Size_2 still determine the quality (and rendering speed)
of the result. You can still get the same artifacts (and now with the
illumination too) if you specify parameters which are too low.

  (I don't mean to say that I think you misunderstood this. I just wanted
to post a note about this, and this felt like a good place to do it.)

  Note that 'adaptive' has no effect on the illumination, only in the
shadows, as previously. This is because there's no easy way to adaptively
sample the illumination, as it is with shadowing. 'adaptive' should,
naturally, still be used for faster rendering because it still speeds
up shadow calculations a lot.

  Basically the end result, when using area_illumination, should be
almost exactly the same as when using a true grid of point lights,
except that it should render considerably faster. (This is because
shadow tests will be much faster.)

  IIRC the 'area_illumination' keyword was suggested by Gilles, and it's
IMO a superb keyword, as it's much more unambiguous than anything containing
the word "full" in it.

-- 
                                                          - Warp

Post a reply to this message