Abstract:

Objective In garment production, pattern-making usually relies on the experience of manual pattern makers, and this process is ofern associated to low efficiency, high cost, and difficulty in meeting large-scale personalized customization demands. Although 3D reverse engineering can generate body-fitting surfaces, it faces challenges in curve deformation and ease allowance control, less conducive to style modifications. Existing parametric methods rely on specialized CAD software, with limited accessibility. In order to address the creation and adjustment of well-fitting patterns, this study proposes a Python-based parametric pattern generation model for automated pattern creation, providing technical support to shorten the production cycle at the pattern-making stage.

Method This study integrates parametric modeling with self-developed biarc curve fitting algorithm to achieve rapid generation and adjustment of well-fitted garment patterns. Taking the generation of a women's shirt pattern in AutoCAD as an example, the research first investigates two types of biarc curves for constructing garment outlines using Python, along with continuity calculations at their connecting points. Next, by combining the prototype method and short-measurement method from garment structural design, body-related parameters are established. The correlations between body dimensions are obtained through anthropometric experiments, and pattern generation/adjustment rules are incorporated to build an automatic pattern generation model. Finally, the automatically generated patterns are evaluated for pressure distribution and fit using CLO 3D software, verifying the feasibility of this generation method.

Results Based on the fitting principles of two types of biarc curves, this study developed a biarc curve algorithm using Python for automatically generating complex curves in garment patterns. The computational results demonstrate that the curves generated by the biarc algorithm possess G⁰ and G¹ continuity, ensuring smooth linearity of the generated curves with high algorithmic accuracy. All connection points exhibit similar G² deviations while demonstrating variations in curve length. A greater t_c value results in a shorter generated curve. Furthermore, when calculating both the deviations at connection points and the curve length of biarc curves constructed using the incenter method, all values were found to be identical to those obtained with t_c=0.5. This further confirms that the connection points of curves drawn via the incenter method are encompassed within the connection point function described. While maintaining low G² values, the algorithm allows for curve length adjustment, enabling convenient processing of complex structural curves in pattern-making. In regional anthropometric experiments, body measurement data were collected from 200 females in Xinjiang, including length dimensions (height, shoulder width, chest width, back width, back length, bust point distance, arm length) and circumference dimens-ions (waist circumference, bust circumference, neck base circumference). Regression analysis was employed to establish the mapping relationships between key body dimensions, which simplified the construction of the automatic pattern generation model. After incorporating adjustment rules, the model was able to carry out rapid modifications of different pattern sections to generate personalized patterns. The automatically generated shirt pattern was imported into virtual software for seam testing. Pressure and fit tests conducted in CLO 3D software reveal reasonable garment pressure distribution and uniform body-garment clearance, validating the feasibility of the proposed parametric pattern generation method.

Conclusion The Python-generated biarc curves demonstrate both superior smoothness and adjustable length properties, enabling efficient handling of complex pattern contours. The proposed method facilitates rapid pattern generation and modification, providing robust technical support for large-scale personalized apparel pattern production. The anthropometric regression model developed for personalized pattern generation was intentionally designed without body type classification constraints. However, subsequent integration of body type classification with the required measurement data could significantly enhance the model's generalization capability and improve pattern accuracy. The proposed method proves universally applicable for automatic generation of diverse garment patterns, with promising potential for implementation in parametric pattern library development.

Key words: garment pattern, women's shirt, parametric pattern, fitted pattern, pattern generation, Python