Pose-Inspired Shape Synthesis and Functional Hybrid.

Area of Science:

• Computer Graphics and Geometric Modeling
• Human-Computer Interaction
• Computational Design

Background:

• Shape synthesis by reusing existing model parts is a growing research area.
• Combining models across different categories for multi-functional object design presents significant challenges due to a lack of natural correspondences.
• Existing methods struggle to bridge the gap between diverse object categories for composite design.

Purpose of the Study:

• To introduce a shape synthesis approach for functional hybrid creation.
• To enable the design of multi-function objects by combining cross-class models.
• To utilize human pose as a mechanism for describing object affordance and facilitating composite design.

Main Methods:

• Introduced human pose to define object affordance, creating a link between different object categories.
• Developed a method to identify candidate shapes that match desired affordances from an input human pose.
• Implemented a recombination process to create well-connected composite models from identified shapes.

Main Results:

• Successfully demonstrated the ability to combine models across different categories for composite design.
• Enabled users to control the design process through pose manipulation and optional category specification.
• Showcased the creation of nontrivial and interesting synthesized models through the proposed approach.

Conclusions:

• The proposed human pose-driven shape synthesis approach effectively addresses the challenge of cross-category model combination.
• This method facilitates intuitive and flexible design of functional hybrid objects.
• The approach is extendable for single or multiple operators and poses, enhancing its practical applicability.