Commercially released cross-platform video games often feature graphical defects, which negatively impact on the reputations of developers and publishers, as well as on the experience of the players. Game industry testing practices often rely on human testers to assure the quality of a game prior to release.

This thesis investigates the question of whether the testing and validation of computer graphics implementations for cross-platform game development can be automated to reduce the burden on human testers, accelerate the testing phase, and improve the quality of games. Using Design Science Research methods and patterns, iterative development and evaluation is undertaken to construct artefacts, drawing upon prior research and industrial works in related fields such as film and television, as well as proprietary game development insight. Elements of existing automated testing systems and image comparison techniques are combined with industry standard cross-platform development tools and methods to create a reusable and generalisable model of an automated test system, featuring image comparison methods, and record and playback techniques for cross-platform game implementations.

In conclusion, this thesis shows that it is possible to create a novel system that is able to detect graphical defects in output from a cross-platform game implementation.