Learning visual balance from large-scale datasets of aesthetically highly rated images

Abstract

The concept of visual balance is innate for humans, and influences how we perceive visual aesthetics and cognize harmony. Although visual balance is a vital principle of design and taught in schools of designs, it is barely quantified. On the other hand, with emergence of automantic/semi-automatic visual designs for self-publishing, learning visual balance and computationally modeling it, may escalate aesthetics of such designs. In this paper, we present how questing for understanding visual balance inspired us to revisit one of the well-known theories in visual arts, the so called theory of "visual rightness", elucidated by Arnheim. We define Arnheim's hypothesis as a design mining problem with the goal of learning visual balance from work of professionals. We collected a dataset of 120K images that are aesthetically highly rated, from a professional photography website. We then computed factors that contribute to visual balance based on the notion of visual saliency. We fitted a mixture of Gaussians to the saliency maps of the images, and obtained the hotspots of the images. Our inferred Gaussians align with Arnheim's hotspots, and confirm his theory. Moreover, the results support the viability of the center of mass, symmetry, as well as the Rule of Thirds in our dataset.

Preference for Well-Balanced Saliency in Details Cropped from Photographs

“…Our study provides experimental support for the notion widely held in computer vision that saliency calculations can be employed in automated cropping procedures to improve the aesthetic outcome (Santella et al, 2006 ; Liu et al, 2010 ; Jahanian et al, 2015 ; Wang et al, 2015 ). In a well-controlled psychological experiment, we demonstrate that the centering of saliency mass onto the geometric image center results in images that are preferred by viewers compared to more unbalanced images from the same photograph.…”

“…It represents the unique point where the weighted distributed mass of the property is equal on either side of an image axis; an image is more balanced if this point is located closer to the geometrical image center. Jahanian et al ( 2015 ) studied pictorial balance by modeling visual weight with the saliency of low-level visual features. For a large set of aesthetically pleasing photographs, they obtained results compatible with Arnheim ( 1954 ) concept of major axes of composition, including the relevance of the image center.…”

Computational and Experimental Approaches to Visual Aesthetics

“…The century-old concept of pictorial balance is related to symmetry, but on a more complex level. Unlike symmetry, it is considered to be an important and universal factor that contributes to the aesthetic appreciation of most types of images, including abstract visual patterns, photographs and artworks (McManus et al, 1985 ; Gershoni and Hochstein, 2011 ; Jahanian et al, 2015 ). According to Arnheim's Gestalt theory of visual balance (Arnheim, 1954 ), an image is balanced if the center of the displayed attractions is placed on any of the major axes of the image (vertical, horizontal and diagonal).…”

Symmetry and Balance as Factors of Aesthetic Appreciation: Ethel Puffer’s (1903) “Studies in Symmetry” Revised

“…Accordingly, the weight of an object can be computed by a simple gray-value integration method, i.e., by integrating the gray values over the object's area (e.g., [14]). This type of perceptual weight has to be distinguished from weights based on higher level properties such as perceptual salience, which is related to visual attention [19,20]. However, irrespective of how weight is determined, we further have to assume that persons are able to perceive the common center of 'mass' of all the perceptual element weights in a picture.…”