A model of visual adaptation has been introduced by Ferwerda et al. in [FPSG96]. This model is based on Ward's contrast based scale factor. The difference is in the functions used for just noticeable differences. Ferwerda et al. use the whole set of different functions for scoptic and photopic vision. In the mesotopic range they use a linear combination of scotopic and photopic functions.
Visual acuity is also taken into account by this model, and dark image parts are simply blurred using a Gaussian convolution filter, since the human visual system is not capable of resolving details in dark areas (see Human Vision section).
The temporal aspect of adaptation is also taken into account. It is possible to simulate an adaptation process over time using this model.
As it is necessary to determine adaptation luminances for using this model, the authors suggest to assign display adaptation to half the maximum screen luminance (same as Ward in his contrast based scale-factor), and the world adaptation to half the maximum of the raw image luminance (different approach than used by Ward). This is probably the weakest point of the whole mapping technique. Imagine, for example, one extremely bright pixel at the edge of the image. It certainly will not influence adaptation level that much, as it will seem from this approach.
This is the first, and still the only mapping method that accounts for temporal change in adaptation. Therefore, if such visualisation is needed, this is the only possible choice of tone mapping. Ward et al. developed a new visibility matching operator, that borrows heavily from Ferwerda's model, but the adaptation level determination is much improved. This improved model will be described in more detail next.