Modeling Aesthetic Preferences in 3D Shapes: A Large-Scale Paired Comparison Study Across Object Categories

TL;DR

This work addresses how humans perceive the aesthetics of 3D shapes and the need for large-scale empirical grounding. It combines a 22,301-pairwise-comparison dataset with a Bradley-Terry latent-score model and non-linear Random Forests analyzed via SHAP to identify interpretable geometric drivers. The study finds both universal principles (e.g., compactness) and category-specific drivers (e.g., proportionality in chairs), with non-linear interactions crucial for accurate prediction. The results yield actionable design guidance across object categories and demonstrate the value of interpretable, data-driven aesthetics modeling for design tools and cognitive understanding.

Abstract

Human aesthetic preferences for 3D shapes are central to industrial design, virtual reality, and consumer product development. However, most computational models of 3D aesthetics lack empirical grounding in large-scale human judgments, limiting their practical relevance. We present a large-scale study of human preferences. We collected 22,301 pairwise comparisons across five object categories (chairs, tables, mugs, lamps, and dining chairs) via Amazon Mechanical Turk. Building on a previously published dataset~\cite{dev2020learning}, we introduce new non-linear modeling and cross-category analysis to uncover the geometric drivers of aesthetic preference. We apply the Bradley-Terry model to infer latent aesthetic scores and use Random Forests with SHAP analysis to identify and interpret the most influential geometric features (e.g., symmetry, curvature, compactness). Our cross-category analysis reveals both universal principles and domain-specific trends in aesthetic preferences. We focus on human interpretable geometric features to ensure model transparency and actionable design insights, rather than relying on black-box deep learning approaches. Our findings bridge computational aesthetics and cognitive science, providing practical guidance for designers and a publicly available dataset to support reproducibility. This work advances the understanding of 3D shape aesthetics through a human-centric, data-driven framework.

Figures (8)

• Figure 1: Representative 3D shapes illustrating high and low values of key geometric features: mean curvature (club chairs), skeleton complexity (dining chairs), and silhouette complexity (lamps). These examples visualize the types of shapes associated with different ends of the aesthetic score distributions in each category.
• Figure 2: Example feature distributions for individual categories. Left: Mean Curvature for the club category. Right: Silhouette Complexity for the lamp category. These plots highlight the variation of key features across shapes within each category.
• Figure 3: Visualized Bradley-Terry aesthetic rankings for four categories: dining chairs, lamps, clubs and mugs. For each category, the top row shows the 10 shapes with the highest inferred aesthetic scores, the middle row shows 10 randomly selected shapes, and the bottom row shows the 10 shapes with the lowest scores. Shapes are ordered from highest (left) to lowest (right) within each row, illustrating the diversity of human preferences and geometric variation.
• Figure 4: Distribution of aesthetic scores: club chairs, dining chairs, lamps, and mugs.
• Figure 5: Pearson correlation heatmaps showing linear correlations between geometric features and inferred BT aesthetic scores for the Dining (top) and Lamp (bottom) categories. Weak correlations highlight the limitations of linear analysis.