UC Irvine

Modeling cloth appearance, an active research topics in computer graphics for decades,generally boils down to representing the geometry and optics of a fabric. Existing models can be broadly classified into two categories. Surface-based models depict cloth geometry using smooth 2D sheets (expressed as polygonal meshes) and optics using specialized reflectance models. Micro-appearance models, on the other hand, express cloth geometry at the microscale down to individual micron-diameter fibers and optics by utilizing volumetric or fiberbased light scattering models. In practice, both categories have their own advantages and disadvantages. The surface-based models are typically light-weight, easy to edit, and can faithfully reproduce the appearance of a fabric at the macro-scale. Thus, they have been widely used in the computer graphics industry. However, these models lack the fine-grained details that are crucial for cloth rendering under close-up views. The micro-appearance models, in contrast, are capable of generating renderings with remarkably high fidelity and details. Unfortunately, as these models are highly data-intensive and usually difficult to manipulate, their practical use has been quite limited. In this dissertation, we present a family of new techniques to efficiently build and render fabric models with fiber-level details. Our first contribution tackles the challenge an extension of the micro-structure models by supporting fabric mechanics. In this light, we introduce a mechanics-aware model that adapts the micro structures of cloth yarns in a physics-based manner. Our second contribution focuses on a practical appearance model for woven fabrics that represents the cloth in ply-level and offer an interactive rendering process. Lastly, our third contribution aims for extending this practical model for knitted fabrics. The approaches presented in this dissertation have created realistic renderings with fiberlevel details. The results match the measurements and offer an accurate appearance. The remaining challenges include developing a more complex representation of cloth while preserving all the fiber-level details and allow transition between different level of details (i.e. surface-based, ply-based, and fiber-based models). We also believe the techniques discussed in this dissertation can inspire insights for other materials beyond fabrics.