[Radiance-general] Radiance and Lumen Micro Differences

[John Mardaljevic]

Martin,

> In general, Radiance is more accurate than other programs if you make
> extensive use of small patches with mkillum, set the -av parameter to a
> reasonable value, and use the following parameters:
> -ab 5
> -ad 2048 or higher
> -as 1024
> -ar 5000 
> -ds .05
 
> If you use -ab 1 instead of 5, your values are typically half of what they
> should be. If you use -ab 3, they might be 30% too low. 
 
That ar setting is rather on the high side.  Also, the high accuracy
that was achieved with the BRE-IDMP dataset relied only on the ambient
calculation - no mkillum.  Some tests I carried out at the time (95-97)
showed that accurate pure ambient tended to be faster than comparably
accurate mkillum.  For both cases the parameters were gradually refined
until the predictions converged to the measured values (method described
in Chap 6 of RwR).  Radiance has changed a bit in the meantime, but I'd
wager that pure ambient is still an effective way to achieve high accuracy.
And maybe even more efficient too.  Perhaps I should repeat the tests using
the latest version - but don't hold your breath just yet.
In any case, I'd stick with the method of testing for convergence (Ch 6 RwR).
The "optimum" parameter combination will depend on the scene, and convergence
testing by progressive refinement is the quickest way to get there.
It's the most reliable method too.

-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)

jm at dmu.ac.uk   
http://www.iesd.dmu.ac.uk/~jm