The 3D curved surface shapes of tight and flared skirts were predicted precisely by the angle curvatures (concentrated Gaussian curvature Kc, concentrated geodesic curvature kc, and concentrated mean curvature Hc), model sizes, skirt designs, and fabrics. All of the 72 skirts, encompassing 3 female body models (mean body sizes of Japanese women in their 20s, 40s, and 70s), 6 kinds of fabrics, tight skirts, and 3 kinds of flared skirts, were investigated with attention to the differences of the curved surface shapes in detail. It has been found that it is possible to predict the number of nodes on hemlines for these 72 skirts with the three totalized feature factors, |Σ±Kc|+|Σ±kc|, |Σ±Hc|, and weight (g/ cm 2) × hemline length (cm), underlying the physical properties of fabric and model sizes, based on slightly high or high correlation coefficients (r = 0.73 to r = 0.93). Therefore, the number of nodes on hemlines of these 72 skirts can be calculated by these three totalized feature factors using a multiple regression analyzing technique, (multiple regression coefficient R = 0.96), based on higher prediction accuracy.

The 72 3D skirt curved shapes, including the tight and 3 kinds of flared skirts, on 6 kinds of fabrics, were categorized by using the both evaluations of both sensitivity images and the physical properties for the fabrics and designs of the skirt. The sensitivity image evaluations by 80 young Japanese women were adopted into the five fabric and the six design curved shape images, while the eight design curved shape parameters and the six fabric curved shape parameters were adopted using the physical properties. Both evaluation values of the 72 skirts were analyzed using the principal component analysis (PCA), and the four PC scores were classified into four skirt curved shape types. Three flared skirt types (9 all hem length flared skirts of thin and light fabric F, 22 high hem length flared skirts, and 23 low hem length flared skirts) are part of the Lady & Feminine Image Style, which uses the different fabric curved shape images and physical property parameter values with the 3D curvatures. While the low hem length type of 18 all tight skirts is the Sharp & Modern Image Style with high tension, hard skirt shape, and the small flare node images that are based mainly on the low 3D curvature values. Furthermore, the predictions of the seven sensitivity images using the measurable physical property parameter values of the skirts were confirmed.