The apparel industry has witnessed a significant shift toward 3D virtual design, driven by advances in simulation and digitization technologies. Central to this transformation are digital twins, which allow designers to test and refine garment designs virtually, reducing reliance on physical prototyping. A critical aspect of this process is accurately capturing the drape behavior of fabrics, as it impacts both the aesthetic and functional properties of garments. Traditional fabric digitization methods involve mechanical testing of fabric samples, but they are labor-intensive, costly, and require specialized equipment. Recently, AI-based digitization techniques have emerged, offering potential advantages in terms of cost and logistics, though their accuracy remains uncertain. This study addresses the gap in knowledge regarding the accuracy of AI-based fabric digitization by utilizing the Cusick drape test-a standardized method for assessing fabric drape behavior. We introduce a reference digitization pipeline based on traditional mechanical testing and perform a comprehensive comparative study of six commercially available digitization methods, including both traditional and AI-based approaches. Using a diverse set of fabrics, we evaluate the accuracy of the digitized fabrics against real-world measurements. Our results demonstrate that AI-based methods can achieve competitive accuracy, while also highlighting areas for further improvement. We present a publicly available dataset containing real fabric properties, their digitized versions, as well as their corresponding real and digital drape metrics, which can serve as a valuable resource for further research and development in fabric digitization.
Sánchez-Banderas et al. (Thu,) studied this question.