With the development of 3D body scanning and 3D virtual garment simulation technologies, online shopping is becoming more promising. However, little research has been done on the fidelity and accuracy of 3D virtual dressing software to prove that it can represent a real person in a garment so that the virtual try-on can be used reliably for apparel fit assessment.
The purpose of the research was (1) to investigate the fidelity and accuracy of a 3D virtual garment simulation tool in assessing fit on a 3D virtual model by comparing the resulting fit on the 3D virtual model and on the participant real body through participant evaluation and (2) to evaluate the effectiveness of the tool for 3D online virtual clothing shopping for consumers. Ferwerda's (2003) framework for functional realism in computer graphics is adapted as a framework for this study.
A 3D virtual garment simulation software package developed by a leading US company was selected. Thirty-seven participants were recruited. Two questionnaires and an interview were used to collect quantitative and qualitative data. A virtual model was made from the participant's scan and virtual pants were simulated on the participant's virtual model. The participants evaluated fit of their selected pants simulation on their virtual model. The participants evaluated the test pants fit on their body, and they compared the test pants fit to the virtual pants fit. The quantitative data from the questionnaires were analyzed using descriptive statistics, two-way repeated measures ANOVA, and independent t-test. The qualitative data from the interview were analyzed using the content analysis technique.
The general participant responses were that the visual information from the virtual model provided them with an overall idea about the pants fit. They were especially impressed that the visual information of overall pants silhouette in relation to their body shape was very accurate. However, the simulation was not completely accurate. The major difference between the fit on the virtual model and the fit on the body was inaccurate material representations. The fabric simulation was laying smoothly on the virtual model, but many wrinkles were visible on the body. Additionally, the fabric texture was inaccurately represented and the simulated fabric was described as resembling a knit fabric.
Technological limitations were found in accurately representing garment to body shape relationships. Because 3D images from a body scanner were solid and uncompressible images, the shape of the body scan images could not be modified in accordance to garment tension.
The results of the study showed that the overall accuracy and fidelity of the virtual simulation tool was moderately good. The results of the study will help software developers in improving the 3D virtual dressing technologies that are accurate enough for consumers to make a satisfactory and reliable online purchase decision.