I was wondering, does the compiler take into account visibility info when doing the light stage? It wouldn’t have to light an object when the light source is in a leaf node that can’t see the object… I think.
Does this mean that doing a fullvis may save on lighting time?
fraco
Ydnar, are you sure about that? I know nothing about the actual implementation of the compiler, but I would have throught that if you create a powerful light source and set the light bounce switch then it would be possible for light to bounce off walls into other leaf nodes not in the player’s line of sight from the current leaf node.
If it actually cuts the light so it doesn’t bounce like this then I would have though you would get some really odd shadows… probably with sharp edges by leaf portals.
Just a thought 
The radiosity algorithm creates new lightsources for every surface (chopped up, actually) that gets lit by the previous pass. Every new unique lightsource has its own envelope and vis data checking.
So, no problem there. 
y
I see. That’s a cunning way to do it :drink:
So for every light bounce point the compiler places a new light source? Does that mean that there are millions of “bounce sources” scattered around the map when running the Lighting phase?