### Confused about Shifted Gamma micro-facet distribution

Posted:

**Mon Oct 13, 2014 8:54 am**Hi,

I've been studying the paper 'Accurate fitting of measured reflectances using a Shifted Gamma micro-facet distribution' a couple of times now:

http://hal.inria.fr/docs/00/70/23/04/PDF/paper.pdf

I stumbled upon a peculiarity today when thinking about measuring data. The paper states that two slices of the BRDF are enough to fit the model to, where Theta D = 0 is one, and Theta H = 0 is the other.

Now, Theta D = 0 is the one that confuses me. My interpretation in this case is that input direction = output direction, so to measure it you would have light source and camera at the same direction relative to the sample. The paper refers to Matusiks data, and looking through his papers, it looks as if light source and camera would be blocking each other for such measurement angles, and for quite a few of the nearby measurements as well.

http://people.csail.mit.edu/wojciech/pubs/phdthesis.pdf

Unless I'm misunderstanding the paper, I'd say that this data for input direction = output direction is interpolated for a number of degrees. My worry here is that one of the two slices stated to be necessary for fitting to the shifted gamma micro-facet model is not actually measured at a high accuracy but a crude approximation! Any thoughts on this? It seems that one would need to use a beam splitter or something to actually measure these angles, so any experience with this topic would be super-interesting to hear about!

I've been studying the paper 'Accurate fitting of measured reflectances using a Shifted Gamma micro-facet distribution' a couple of times now:

http://hal.inria.fr/docs/00/70/23/04/PDF/paper.pdf

I stumbled upon a peculiarity today when thinking about measuring data. The paper states that two slices of the BRDF are enough to fit the model to, where Theta D = 0 is one, and Theta H = 0 is the other.

Now, Theta D = 0 is the one that confuses me. My interpretation in this case is that input direction = output direction, so to measure it you would have light source and camera at the same direction relative to the sample. The paper refers to Matusiks data, and looking through his papers, it looks as if light source and camera would be blocking each other for such measurement angles, and for quite a few of the nearby measurements as well.

http://people.csail.mit.edu/wojciech/pubs/phdthesis.pdf

Unless I'm misunderstanding the paper, I'd say that this data for input direction = output direction is interpolated for a number of degrees. My worry here is that one of the two slices stated to be necessary for fitting to the shifted gamma micro-facet model is not actually measured at a high accuracy but a crude approximation! Any thoughts on this? It seems that one would need to use a beam splitter or something to actually measure these angles, so any experience with this topic would be super-interesting to hear about!