Research

This project began as an endeavor to explore concept art in serious games for the Humaginarium project. Diabetes Agonistes is a proposed video game that aims to help enable adults to gain control over their chronic illness. Serious games are games designed to entertain as well as serve a beneficial purpose like education or simulation training. They often are lacking in resources to create robust, engaging experiences seen in AAA games.

Research was done on the topic of art styles and how graphics could influence this problem experienced in serious games. Interestingly, there was not much success in finding peer reviewed literature on art styles. Studies that did exist were conflicting or were vague in their definitions of terms such as realism. The initial goal was to compare how different art styles in this realism continuum might affect learning and how this correlates to user preferences. It became clear, however, that without an instrument to quantify the art, a proper comparison would be difficult. It was at this point that my research took a new direction; taking a few steps back, I began laying out a framework that would allow something like this to be accomplished in the future. It appeared that no known framework was made yet that put art on a scale by quantifying attributes of visual art.

Deliverables

Instrument

This project aimed to see if a valid instrument could be created to quantify elements of visual art. The instrument was created by looking at illustrations, listing the attributes that made illustrations more realistic or more abstract, and categorizing those attributes based upon the principles of visual art. A Likert scale scoring system was implemented to account for the nuances within each attribute. For example, two illustrations may both have outlines, but the difference in line weight and color could make one more abstract looking over the other. 

Random artwork in varying art styles and perceived levels on the realism continuum were put through the test in order to see if the scores they received reflected their subjective ranking. This process was repeated, and questions were fine-tuned until reaching the framework of the initial instrument. 

Concept Art

Color, rendering, and proportion were the categories of focus for testing due to their initially perceived importance in influencing the realism scale. Before creating the concept art, values were assigned for all three categories per concept art piece. The six images’ scores evenly spanned along the range of the instrument in order to have sufficient representation of the continuum. After creating point values, qualitative attributes for each piece were assigned to reflect this score. 

Thumbnail sketches were created that reflected a structure relating to type 2 diabetes.

Refinement of the rough sketches as well as color comps were created to flesh out each piece. This helped keep the range of artwork distributed across the spectrum. 

3D modeling programs, such as ZBrush and 3ds Max, were used for the 3D rendered images. The remaining illustrations were created in Adobe Photoshop and Adobe Fresco. 

Results

A committee meeting was held to discuss the comprehensiveness of the instrument. After further revisions, the new version of the instrument was then sent out to the three committee members to individually evaluate the concept art pieces. Committee members were asked to first rank the artwork based upon their initial judgment of where they fall on the continuum in relation to one another. They then evaluated each piece of artwork using the instrument. The scores from the instrument were then compared to the initial rankings.

The images were placed on the continuum based upon the scores they received. The data was also organized to show the initial ranking of the artwork against the instrument scores. The scores of the images closely reflected the initial scores given prior to running them through.

A range in values between the highest and lowest scores of the content validity members was determined. The highest range of the six images was .39, which suggests a relatively high precision rate between the individual rankings. The low range between responses as well as placement comparison suggests a pattern and potential for an objective, quantifiable score. 

The ability to quantitatively evaluate a single continuum might pave the way for future investigation, leading to potentially powerful insight for medical illustrators and game developers alike.

This study suggests that a valid instrument can be developed to measure the level of realism. Translating this qualitative data into quantifiable values will give structure and validity to future research. With more testing and refinement, a valid instrument could be created to objectively measure the realism-abstraction continuum.