We estimate the lighting parameter from photograph and initial 3D geomtery.
we assume that the object is perfact diffuse (Lambertian), this allow us to represent the radiance with
only the point normal, thus it's possible to use spherical harmonic to estimate the lighting.

/**************************************************************************************************************/
the second light esimator is based on Oren-nayar at it's simple form where we assume that
the illumination is based on direct lighting and the source direction is known and it is equal to
camera view direction. the estimated light here will be one value notice that the roughness will be one value
*****************************************************************************************************************/
Results : https://www.youtube.com/watch?v=n7nWLNQ3ius

PLease refrence this code by this:

@article{saouli2018accurate,
  title={Accurate, dense and shading-aware multi-view stereo reconstruction using metaheuritic optimization},
  author={Saouli, Abdelhak and Babahenini, Mohamed Chaouki and Medjram, Sofiane},
  journal={Multimedia Tools and Applications},
  pages={1--25},
  year={2018},
  publisher={Springer}
}