Low priority (does anyone else use rad, if not, you should!), but I am curious if the rad program has been updated to incorporate any of the recent (# of years?) changes to the ambient calculation. I am thinking most notably about the update to -aw prompted by this thread:
and am curious if the -aw 1 -av 0 0 0 setting is also appropriate for daylight scenes (with a view to bright objects outside).
As an example, given this rif file snippet:
rpict= rtpict -n 8
RESOLUTION= 1000
ZONE= I 0 20 0 20 0 20
QUALITY= H
DETAIL= H
EXPOSURE= 1
UP= Z
INDIRECT= 6
VARIABILITY= H
REPORT= 1
I am also curious about the very low (for an ambient cache run) -lw, the fact that -av doesn’t change with increasing -ab, and that -ab is entirely manual per the INDIRECT variable (could be connected to VARIABILITY?)
This leaves the overture calculation to rtpict, which it handles more efficiently in its own way, which I discussed in my talk at this year’s workshop.
The -lw setting with irradiance caching isn’t all that critical. It just needs to be small enough that you don’t miss deep-tree specular contributions. It becomes more important with -aa 0 (irradiance caching off) when it controls the depth where path tracing takes over.
Regarding the -aw setting working better than it used to (also covered in my talk), I have not been brave enough to change the defaults in rad to enable this feature. You are welcome to give it a go.
The -av setting should be the same no matter what -ab is set to. It is an approximation of the amount of light bouncing around in an equilibrium state. It’s not strictly a “remainder” in that sense.
Thanks, I’m running a pretty complex scene right now with some very high -aa 0 settings and ambient caching using the -av 0 0 0 -aw 1 and fewer bounces, so I will let you know how the new -aw performs in this case. I am testing out some image reconstruction with raytraverse and also want to see if ambient cache + raytraverse offer any compounding efficiencies.
about the overture, when I call rad in this way I get the overture using rtpict (first line in op), but you are saying it is more efficient to skip the first line and just call rtpict (second line only)?
The -aa 0 option turns off irradiance caching, which also turns off the effect of -aw, since it has no cached values upon which to base its estimate.
Regarding rtpict, it doesn’t hurt to run an additional ambient calc, but it isn’t necessary as the ray samples get randomized, avoiding the usual “cache until it hurts” artifacts you get from a straight rpict run.