Fig 1.
Using eye-tracking simulation and spatial analysis to explain people’s preferences of alpine landscapes, which were collected via surveys based on panoramic landscape photographs.
Photographs from Schirpke et al. [11], published under the Creative Commons CC BY 4.0 license.
Fig 2.
Examples of photographs representing 19 different LULC types.
Photographs from Schirpke et al. [11], published under the Creative Commons CC BY 4.0 license.
Fig 3.
Output of the 3M-VAS software.
(1) Original image, (2) Heatmap indicating the probability that areas are seen within the first 3–5 seconds, (3) Hotspots derived from the heatmap, specifying the probability that a person will look somewhere within the hotspot areas within the first 3–5 s), (4): Gaze sequence (most probable viewing order) of the 4 most-likely seen gaze locations, (5) Visual elements (edges, intensity, red-green color contrast, blue-yellow color contrast, faces), indicating how each of the elements contributes to the overall probability. Photograph by E Tasser.
Table 1.
Variables extracted from the hotspots that were identified by 3M-VAS.
Table 2.
Variables extracted from visual photo content analyses.
Fig 4.
Mean preference values (, n = 4–5) of different LULC types, ranging from 1 = ‘least preferred’ to 10 = ‘most preferred’ (Likert scale).
Fig 5.
Examples of identified eye-tracking hotspots by 3M-VAS and the frequency of photos with at least one specific content within the top-hotspots (HPmax) and all hotspots.
Frequencies were calculated considering only those photos containing the specific contents, e.g., number of hotspots with water in relation to all photos including hotspots with water. Photographs from Schirpke et al. [11], published under the Creative Commons CC BY 4.0 license.
Fig 6.
Contribution of visual elements (edges, intensity, red-green color contrast, blue-yellow color contrast, faces) to the probability that hotspot areas are seen within the first 3–5 seconds across all LULC types (a), urban LULC types (b), agricultural LULC types (c), forest types (d), alpine LULC types (e) and water LULC types (f).
Table 3.
Result of the backward stepwise linear regression, including only eye-tracking predictors with tolerance >0.1 and variance inflation factor (VIF) <10 during collinearity diagnostics.
Only variables with p<0.1 are shown.
Table 4.
Result of the backward stepwise linear regression, including landscape metrics, photo content indictors and hotspot predictors with tolerance >0.1 and variance inflation factor (VIF) <10 during collinearity diagnostics.
Only variables with p<0.1 are shown.