Fig 1.
Schematic illustrations of the five blur treatments applied to each of the six photographs of full-scale railway scenes.
A sixth condition involved no blur treatment. The red line indicates the contour of focus (the precise position of this varied across the six different scenes). The grey levels (given by the bar in the lower right corner) and blue lines indicate the variation of Gaussian blur across the image. In (a, b, d) the blur was constant horizontally and varied vertically. In (c) the blur was constant vertically and varied horizontally. In (e) the blur was uniform across the entire image. See text for further details. The treatment in (a) is sometimes referred to as fake tilt-shift miniaturization.
Fig 2.
(a-e) The five image treatments described in Fig 1 applied to one of the images of a full-scale railway scene. (f) The original image (photograph by Don Burgess). See the Acknowledgements section for a link to the full set of stimuli used in the experiment.
Fig 3.
The six scale model images.
Fig 4.
The application of blur treatment from Fig 1A (sometimes known as fake tilt-shift miniaturization) to each of the six full-scale railway scenes.
The original images were published under a creative commons license by the following authors: (a) Don Burgess, (b) Alan Murray-Rust, (c) Andy F, (d) Ingy the Wingy, (e) Phil Sangwell, (f) Roger Geach.
Fig 5.
Experimental results: The percentage of trials in which the full-scale scene was correctly identified.
Error bars show 95% confidence limits on the mean triplet scores (n = 36) determined by bootstrapping. The double asterisks (**) indicate the conditions that were significantly different from chance (see text for statistical details). Responses in the gradient blur condition (green bar) were significantly less often correct than each of the other conditions except for the square wave blur condition (purple bar). See text for details.