How realistic are painted lightnings? Quantitative comparison of the morphology of painted and real lightnings: a psychophysical approach.

Inspired by the pioneer work of the nineteenth century photographer, William Nicholson Jennings, we studied quantitatively how realistic painted lightnings are. In order to answer this question, we examined 100 paintings and 400 photographs of lightnings. We used our software package to process and evaluate the morphology of lightnings. Three morphological parameters of the main lightning branch were analysed: (i) number of branches Nb, (ii) relative length r, and (iii) number of local maxima (peaks) Np of the turning angle distribution. We concluded: (i) Painted lightnings differ from real ones in Nb and Np. (ii) The r-values of painted and real lightnings vary in the same range. (iii) 67 and 22% of the studied painted and real lightnings were non-bifurcating (Nb = 1, meaning only the main branch), the maximum of Nb of painted and real lightnings is 11 and 51, respectively, and painted bifurcating lightnings possess mostly 2-4 branches, while real lightnings have mostly 2-10 branches. To understand these findings, we performed two psychophysical experiments with 10 test persons, whose task was to guess Nb on photographs of real lightnings which were flashed for short time periods Δt = 0.5, 0.75 and 1 s (characteristic to lightnings) on a monitor. We obtained that (i) test persons can estimate the number of lightning branches quite correctly if Nb ≤ 11. (ii) If Nb > 11, its value is strongly underestimated with exponentially increasing difference between the real and estimated numbers. (iii) The estimation is independent of the flashing period Δt of lightning photos/pictures. (iv) The estimation is more accurate, if skeletonized lightning pictures are flashed, rather than real lightning photos. These findings explain why artists usually illustrate lightnings with branches not larger than 11.