How close home-made headlight beamshots can match real goniometer measurements? A very brief description of the method: simulations and comments.
Data were normalized to approximately match the 80lux TA isoline size. For the purpose of the match (the 4 images on this page): values of isolines (e.g., ..,80,70,60,..), headlight height (0.7m), aim distance (10m) - are as Olaf Schultz uses. Isolines not colored. Colormap is (almost) same.
Also note: there are 2 mistakes in Olaf Schultz's gnuplot script, see comments in the source: sources-octave/main_vis.m for details. I'm reproducing the mistakes here.
A lot to observe.. help yourself.
My older colorimeter normalization gives only 5% higher values - very good.
However, I think that the above reasons can only have little influence on the discrepancy in the cutoff area. May be it's not the very same headlight sample, and they simply are somewhat different.