Sphere-less method feasibility?

[jacopomazzoni]

Someone pointed me to this paper:

https://ore.exeter.ac.uk/repository/handle/10871/40326

Is any of this possible to implement in the software?
The claims of improved quality results and the removal of the spheres requirement seem really interesting.

Best,
Jacopo

[ponchio]

I know this paper. The main problem is that the paper is quite vague on the implementation.
Another problem is that the method seems to be based on 3 heuristic functions and do not have a solid mathematical theory (or at least is not clear from the text).

I have the distinct impression that this method works only if the surface is flat and mostly lambertian, BUT, since this is a common case, it might be worth to try. I quickly made up a test (but with different math) and got some non terrible result. I will try to see if I can create something convincing.

Thanks.

[jacopomazzoni]

Wonderful!

I was worried about the centroid method for hilight detection, it would fail horribly on textured surfaces with white areas.

I was also curious about the light intensity compensation method because that's been a bit of an issue with BTF RGBW light strips being inconsistent.

Thank you as always!