Image-Based Reverse Engineering and Visual Prototyping of Woven Cloth.

Area of Science:

• Computer Graphics
• Textile Research

Background:

• Realistic cloth visualization is crucial for computer graphics and computer-aided design.
• Existing methods for cloth modeling are often difficult to edit or require extensive data capture.

Purpose of the Study:

• To develop a pipeline for reverse engineering cloth models from single images.
• To create a parametrized cloth model that is easily editable.

Main Methods:

• Proposed a geometric yarn model integrating textile research.
• Developed an automatic analysis approach to estimate yarn paths, widths, variations, and weave patterns from an image.

Main Results:

• Successfully modeled the appearance of original cloth samples from single images.
• The derived properties provide a physically plausible and editable basis for cloth models.

Conclusions:

• The proposed pipeline enables efficient and editable cloth modeling for computer-aided design.
• This approach offers a practical solution for capturing cloth variations from limited data.