OVERVIEW
STATEMENT OF THE PROBLEM
TARGET AUDIENCE
RESULT
To create CC systems that are capable of generating help-ful GD solutions, one must first be able to create objectivemetrics to describe the quality of the expected designs. Nev-ertheless, due to the subjectivity of GD aesthetics, creatingcapable metrics for evaluating GD is still an open problem.GD evaluation metrics may focus, for example, on theconcept of the work, the legibility of the contents, innovationdegree and other even more subjective features such as thepersonal taste of the target public. Another relevant featureis visual balance, which often relates to the visual weight ofthe items in the composition, on each side of a given axis.
Being able to evaluate aesthetics automatically is oneof the fundamental needs for creating robust and au-tonomous computational creativity systems. In GraphicDesign (G D), many aesthetic features might need tobe considered simultaneously to properly evaluate GDartefacts, e.g. their visual relation to the concept ofthe work, legibility, innovation degree and the personaltaste of the target public. Another relevant feature isthe balance of the elements in the composition. Thispaper presents and tests an approach for evaluating thepage balance of GD posters. Furthermore, it comparesthe evaluation computed by the developed method withthe evaluation made manually by graphic designers andother creative practitioners.
To test the presented approach, a set of 120 G D posterscreated by different authors and gathered from varioussources were evaluated manually by graphic designers andCC practitioners, by means of a user survey.
The results suggested the proposed method could match,at around 80%, the balance evaluation made by the respon-dents. Moreover, the results indicated a possible correlationbetween page balance and visual pleasantness, at least, forthe current experimental setup.Future work must focus on testing the proposed approachas a fitness assignment method for an automatic evolution-ary system. As assessing balance alone may be reductiveto evaluate GD posters, we must complement it with othermetrics, such as for assessing legibility or innovation de-gree. Lastly, we must further study how different personalbackgrounds or additional visual features, such as apparentmovement, hue and saturation, may impact the calculationof visual balance.
Towards the Automatic Evaluation of Visual Balance for Graphic Design Posters
Author: Daniel Lopes, João Nuno Correia, & Penousal Machado
Date: June, 2023