Skip to main content
. 2019 Apr 13;10(2):2041669519842895. doi: 10.1177/2041669519842895

Figure 1.

Figure 1.

(a) A groomed ski run. A patch of groomed snow in sunlight (b) and in shade (c) and their image spectra (d). Photos in (b) and (c) were taken minutes apart on the same slope and day with an iPhone X. Images were converted to greyscale, assumed to have an encoding gamma of 2, scaled to be in the range [−1 1], and normalized so that the mean pixel intensity was zero. Finally, I applied a circular aperture with a cosine edge to minimize wrapping artifacts during analysis. (d) The dashed red line and solid blue line show normalized contrast energy for the sunlit and shaded patch, respectively, as a function of spatial frequency. Contrast energy was calculated in the frequency domain, summing across orientations and normalized to the maximum contrast across images.