There may be an effect in addition to those already noted.
Just for the record, if I make a model from scratch using
textfiles I don't have any of these problems. Additionally, the
problem crops up whether I'm using Radiance on Cygwin, OS X,
or Mandrake Linux, so it's obviously something I'm doing or
a problem with the models coming out of Revit.
There have been a number of instances when I've been provided
with CAD models (usually in 3DS format) containing perturbed
surface normals. The perturbation applied in the CAD models
is, in my experience, always a cosmetic hack. For example, to
give variation in shading across a flat facade - here the
building was a tall block with a square cross-section but the
surface normals were perturbed as if it was a cylinder.
These perturbations are preserved in the conversion from
3DS -> mgf -> rad and may go un-noticed in later renderings.
At least I didn't notice them at first. They are more likely
to reveal themselves in illumination/irradiation images.
Might they also bamboozle the ambient interpolation algorithms (Greg?).
The offending polygons may appear something like this in the rad file:
mat0 texfunc T-nor
4 dx dy dz tmesh.cal
0
10 1
-0.00645062 0.00858814 -51.95344592
0.00910556 -0.00762344 42.56315834
0.01317944 -0.01247687 71.46587731
T-nor polygon Obj_000001.t1
0
0
9
6483.0762 -6394.1626 725.7263
6463.3491 -6344.1626 681.5585
6502.5605 -6394.1626 699.0936
I strip them out using scripts. Intrestingly, the people who
generated the CAD didn't seem to be aware that they were introducing
these perturbations. Perhaps it happens when they select some
shading option in 3DStudio (or similar).
Any one else encountered these?
-John
···
-----------------------------------------------
Dr. John Mardaljevic
Senior Research Fellow
Institute of Energy and Sustainable Development
De Montfort University
The Gateway
Leicester
LE1 9BH, UK
+44 (0) 116 257 7972
+44 (0) 116 257 7981 (fax)