Speed of Light: Photometric Point Lights, one vs many?
I'm working on converting a model for iray. A long hallway...
The original was designed to be lit by long ceiling panels set as emmissives. I've decided I wish to have a photometric version, as all the emmisives from the ceiling are a bit slow. (I haven't checked to see how the polygons in the ceiling pannels are...if there's many or a few...but I'm going to assume many for the purpose of this question.).
I HAVE been playing with shaping photometric point lights into columns/rods for use in place of emissive lights...basically like flourescent tubes..they're working pretty well so far.
THE QUESTION I'm heading to, though, is it better/quicker, to have for just the single point light streached the full length of the hall way. ie:
-------------------------------------------------------------------------------------------------------------------
OR instead use several smaller length lights instead?
--------------------- --------------------- --------------------- --------------------- ------------------
Comments
How are you changing geometry on point lights? I don't believe you can actually change geometry. Here are 3 examples. First one Iray pt light 10x10 default lumens. Second one 1 x 100 default lumens. Third one 1 x 1000. Cubes are 1 m. No difference
I'm stretching them with scale. It works. I'm doing some render tests right now, so i can post pics and stuff. Make sure you change the geometry of the point light into something else..like a column or plane.
Hmm.. Did 3 tests, not of the broken vs single lights, but to see speed differences between the mesh lights and the point lights (and a 3x size of the point light that I'd shrink in post).
With this test, the mesh lights, not the photometric is about 10% quicker.
(I have no idea if this test is meaningful...if the progression is linear, if it changes over a greater amount of time...)
EDIT: Did a series of tests with Mesh lights at 3x the size, to see if i could track how these progress.
Well.. finding all sorts of data that I don't know what all it means, but to answer my original question, I did a 3x size render with the photometric light, then split it into 3, and did it again..the single light had slightly more iterations and convergence in a full 2 hour run.