Spot The Difference: AI, Please Make This for Me!

Abstract

Play and games are an inseparable part of our lives, and we have been playing various games from our childhood to adulthood. One of the widely recognized games is "Spot-the-Difference", which has been played and employed in a variety of domains, such as education, training, and entertainment. However, designing a custom "Spot-the-Difference"game for a certain domain requires consideration in terms of the level of game difficulty, game interest, and human intervention, therefore, is non-trivial. In this paper, we propose a novel framework based on Human-AI collaboration to automatically generate images for the "Spot-the-Difference"game. We employ Maskformer and Inpainting Stable Diffusion to automatically identify and re-draw a set of regions in the image. Finally, we conducted a user study with 19 participants to evaluate our framework. From the experimental result, we found that AI-generated game images were enjoyable, but there is still room for improvement in the quality of the overall game playing.