Parametric prediction of garment hood silhouette for simulation‑efficient virtual prototyping: from pattern and material to CAD forecast

Virtual prototyping of clothes relies on iterative trial‑and‑error simulation when searching for a desired silhouette, which wastes time, energy and computing resources. Hoods are a case in point: their draped form depends on complex interactions among construction, material and the wearer’s head and therefore cannot be predicted from the pattern alone. This study proposes a predictive framework built on a test stand derived from ISO anthropometric data and 560 simulations performed by systematically varying six construction parameters of a two‑seam hood across 20 plain‑weave virtual fabrics. A radial parameterisation captured the silhouette with 32 parameters in front, side and top views, yielding a structured database of 17,920 values. Regression models using construction dimensions, density and bending stiffness predicted these parameters and were embedded in a ready-to-use parametric CAD drawing for instantaneous graphical forecasting. Physical validation with a 3D‑printed stand and real fabrics showed that 87 % of the observed contour length fell within the prediction band. A probabilistic model demonstrated that the approach can reduce computation time by 90.5-99.6 %, offering substantial resource savings in digital garment development.