Computational production of colour harmony. Part 2: Experimental evaluation of a tool for gui colour scheme creation. Issue 3 (1st November 2011)

• Record Type: Journal Article
• Authors: Moretti, Giovanni; Marsland, Stephen; Lyons, Paul
• Is Part Of: Color research & application. Volume 38:Issue 3(2013:Jun.)
• Page Start: 218
• Page End: 228
• Publication Date: 2011-11-01
• Subjects: Color -- Periodicals; 535
• DOI: 10.1002/col.20735
• Languages: English
• ISSNs: 0361-2317

Abstract

<abstract abstract-type="main"> <title> <x xml:space="preserve">Abstract</x> </title> <p>Although webpage and computer interface designers generally have little experience at generating effective colour schemes, colour selection appears only rarely in user interface design literature. This article describes the experimental evaluation of an algorithmic technique that applies colour harmony rules to the selection of colour schemes for computer interfaces and web pages. The technique uses a genetic algorithm to evolve colour schemes; the evolutionary path is determined by a quantitative colour harmony evaluation function.</p> <p>Our technique first creates abstract colour schemes by applying those rules to specific features of the interface or web page; the user then holistically modifies the scheme's overall colour cast, overall saturation, and light–dark distribution, producing colourings that are both harmonious and usable. We demonstrate experimentally that the software is relatively simple to use and produces colourings that are well‐received by humans.</p> <p>In an earlier article, the criteria for a colour harmony tool for computer interfaces and websites were described and used in the design of the Colour Harmoniser, our software implementation of a system that is based on classical rules of colour harmony, adapted and extended to suit graphical user interfaces.</p> <p>In this article, we describe two sets of experiments that have demonstrated the usability and effectiveness of the Colour Harmoniser tool, compared with standard methods of colour selection. These experiments suggest that the tool functions somewhat more effectively than we originally anticipated, producing colour schemes that were rated more highly on several quality scales than those produced by random choice, by humans who self‐classify as nonartists, and by humans who self‐classify as artists. © 2011 Wiley Periodicals, Inc. Col Res Appl, 38, 218–228, 2013.</p> </abstract>