In serious game development, graphic design needs to be eye-catching, while also depicting subject matter content in a responsible, accurate, and clear way. Previous research has shown that abstract and symbolic game visuals seem to be preferable for learning and providing an engaging experience. Our research focuses on describing the challenges involved in creating effective visual communication through game graphics in cross cultures. In particular, we’re interested in examining if certain styles of visual communication are more or less effective between different cultural demographics. To examine this, we have created a serious game which aims to promote healthy food and nutrition habits to teenagers in both Nepalese and Swedish schools and by doing so also motivate behavioral changes toward healthier eating habits. We are currently conducting studies to see whether preferences and image recognition differ between the two demographical spheres. This paper will only discuss the exploratory study done in Nepal. Ultimately, this paper aims to contribute development guidelines that can aid developers in creating more effective visual communication in their serious games, and we primarily focus on exploring what we call the compromise of ‘clarity’ and ‘appeal’ in the creation of game graphics. We present an initial model for choosing at what level in terms of realism/abstraction and taxonomic hierarchy the graphical components of serious games optimally should be produced in order to solve the dilemma of precise, unmistakable, yet appealing visuals in serious games. It all comes down to two primary decisions: defining the taxonomic hierarchy of the items to depict, and choosing the style in which to depict them. With a better understanding of when different game visuals are more or less appropriate, both in terms of style and in which objects are represented, game developers will be able to balance production costs better while also creating something that strikes the compromise between clarity and appeal.