Smart Grading

Abstract

We present fast and automatic method for garment grading. In the apparel industry, garment is designed to fit standard body, and then it is modified to fit specific body. This modification is called grading. Grading is very specialized and complex work. Therefore, it is extremely time consuming to do grading, also grading is a hard task without an exclusive knowledge. Nowadays, there are in need of the grading techniques in the animation and game productions, since costume design takes an important component in the process. Moreover, the variation of the bodies appearing is broader than the real clothing production, sometime such bodies are difficult to cover with the conventional grading methods. To solve above problems, we introduced retargeting technique which is widely used in the computer graphics field. To use retargeting technique, we need the mediator and the correspondence function. For the mediator of our method, we got the insight from the process of drawing the pattern-making draft. We call this mediator Parameterized draft. Local coordinates systems are good methods for making correspondence. Among others, the mean value coordinates system (MVC) would be an excellent choice, but needs to be improved so that the weights would have positive values. We improved the MVC and call it the omitted mean value coordinates (OMVC). To put it pithily, the mediator is parameterized draft, and correspondence function is the OMVC in our approach. We call this approach smart grading. Smart grading is less time-consuming and easy to implement. Therefore, our approach can minimize designer's specialized know-how and save performing time for the grading of real garment and virtual garment.