TL;DR
This paper introduces an interactive design tool that allows users to create complex 3D surfaces by interlocking simple quadrilateral elements. The system focuses on finding the right "combinatorial geometry" – how elements connect and orient – to achieve a desired shape, making the design process more intuitive and efficient than traditional methods.
What Problem Does This Paper Address?
Designing complex physical objects from simple, interlocking components is a powerful concept, seen in everything from puzzle lamps to architectural structures. However, manually figuring out how to connect and orient thousands of individual pieces to form a specific free-form surface is incredibly challenging and time-consuming. This paper addresses the gap in intuitive design tools for this type of construction, where the final geometry is dictated by the discrete arrangement (topology and orientation) of repeating elements, rather than continuous modeling. Existing methods often focus on generating the interlocking elements themselves, but this work focuses on designing the assemblies using a fixed element type.
Methods
The core of the paper is a forward modeling tool that allows users to interactively explore the space of feasible designs. The approach leverages two fundamental operations:
- Merging: This operation combines two existing shapes by finding compatible interfaces (edge loops) on each shape that can be joined. The system uses geometric and combinatorial techniques to quickly rule out incompatible interfaces and find a realizable embedding.
- Extrusion: This operation allows a user to "sweep" a shape along a path, effectively adding geometry. The tool uses a library of "brush" shapes and an "extrusion apex" on the target mesh to find candidate interfaces for merging.
To handle the complexity, the system represents interlocking elements using a tagged half-edge data structure. It employs an energetic formulation to model the kinematic constraints of the interlocking elements, penalizing stretching and compression to determine the equilibrium shape. For interactive feedback, they also compute discrete Gaussian curvature to predict local shape and feasibility. Crucially, the system generates assembly instructions to guide users through the physical construction process.
Datasets
Not discussed in the paper. The paper focuses on a generative design tool and demonstrates its capabilities with examples created by the authors.
Key Insights and Ideas
- Combinatorial Geometry is Key: The paper highlights that for assemblies made of a single, repeating element, the entire geometry is governed by the element's connectivity (topology) and binary orientation. This shifts the design focus from continuous shape manipulation to discrete arrangement.
- Intuitive Operations for Discrete Design: Merging and extrusion, familiar concepts from traditional modeling, are adapted to work with discrete interlocking elements. This makes the design process more accessible to users who might not be experts in combinatorial geometry.
- Bridging Digital Design and Physical Fabrication: The system not only allows for digital design but also generates assembly instructions, directly addressing the practical challenges of building these complex structures in the real world.
- The Power of Abstraction: By using a library of base shapes and simple operations, the tool abstracts away much of the underlying combinatorial complexity, allowing users to focus on the creative aspect of form-finding.
Key Results
The paper demonstrates the versatility of their approach by creating a diverse set of digital and physical examples, including a chandelier, a dancer, an elephant, and a snake. They report that the design time for these complex shapes was significantly faster (10-25 minutes) than traditional trial-and-error methods, and the fabricated physical models closely matched the predicted digital designs. Assembly times for fabricated models ranged from 155 to 288 minutes, showing a clear advantage over the design phase.
Limitations and Future Work
The authors acknowledge several limitations and suggest future directions:
- Simplifying Kinematic Assumptions: The current model uses simplifying assumptions for interactive rates. More accurate results could be achieved with nonlinear thin-shell and contact mechanics, though this would increase computation time.
- Inverse Design: Exploring an inverse approach where users specify a target geometry to be approximated by interlocking elements is a promising avenue. This would require adapting quad remeshing algorithms and developing ways to abstract input geometry for the restrictive design space.
- More Complex Elements: The current work focuses on a single, quadrilateral element. Future work could explore designs with more complex or varied element types.
- Enhanced Assembly Instructions: While the current assembly instructions are helpful, further refinements could improve the user experience.
Summary updated