| Petr Kellnhofer1 | Tobias Ritschel1,2 | Karol Myszkowski2 | Elmar Eisemann3 | Hans-Peter Seidel1 |
| 1MPI Informatik | 2Saarland University | 3Delft University of Technology |
This is a supplemental for a EGSR 2015 paper "Modeling Luminance Perception at Absolute Threshold". It provides a detail comparison of different dark noise approaches applied to several CG and real world images.
Notice how various types of noise look behave differently on bright and dark background. E.g. the gaussian being purely additive has the same magnitude regardless of the image luminance. Our approach, on the other hand, generates more noise in very dark regions where little photons are available to the eye and the spontaneous rod events may prevail. The noise in brighter region then slovly disappears as the number of arriving photons become sufficiently large (hundreds per second and rod).



