pmap, mkillum and photon ports

Hi all,

I just want to share my experiences with the pmap for radiance 3.7p2, the use of photonports and mkillum. Maybe it is interesting to someone here, and maybe I even get some comment on what I am doing :wink:

First, to introduce, the problem with the pmap is that, for large angle sources, it is only performing well when used with photon ports. The drawback of photonports is, that no interreflection calculation happens outside the port. That means that, when modeling a simple room with a window, only unobstructed parts of the visible environment contribute - buildings, ground planes etc. do not contribute any reflected illuminance.

To solve this, I played a bit with the combination of pmap and illum surfaces. Carsten did that before, but with a different scene, different set-up and different goals. To get the indirect component of light flux into my photon ports, I have to place my illums outside the ports, so that they become (unobstructed) light sources accounting for the light reflected from outside geometry.

I have set up a simple scene as follow: a room 10mx10mx5m, with its floor level at +5m. In the floor of that room, a hole of 5mx5m (pointing DOWN!). At -1m, a ground geometry (disk, diameter 50m). A photon port in the 5mx5m hole. And five mkillum-surfaces, again 10mx10mx5m, so the geometry of the room mirrored over the rooms floor. This means that the bottom mkillum surface is located at +-0m, just 1m above the the ground disk. This bottom surface is generated by gensurf, and is actually an array of 20x20=400 squares of .5mx.5m each.

I place a view at 0,0,5.1 (10cm above the photon-port), pointing upwards, using vth-fisheye view type with 180 degree vertical and horizontal. And now I render four images and sum up all pixel values for each image:

scene classic (no illum, no pmap, ab 10): 1180668
scene with pmap (no illums, but also no photonport): 1013533
scene with illums (no pmap): 1522379
scene with illums and pmap: 1695772

mkillum was started with default values and -ab 10. Rendering with photonport and without mkillum does not really work, as the port does not see a light source.

So for now, the result is that, as long as mkillum is involved, the results are pretty stable. I would not call the two results close, but at least they have similar scales :wink: Having the two pixel value sums for the runs without mkillum being that much lower, I got worried however. Would I have to set much higher ambient parameters, or is there something wrong in my set-up? Is the use of exterior mkillum-koffers allowed?

Cheers, Lars.