A Multi-scale Yarn Appearance Model with Fiber Details

TL;DR

This work addresses the challenge of rendering realistic cloth by replacing explicit sub-yarn geometry with a yarn-curve aggregation and implicit ply/fiber details inside a Bidirectional Yarn Scattering Distribution Function (BYSDF). It introduces a multi-scale shading framework that transitions from near-field to far-field rendering using pixel-coverage integration and an attenuation term based on Beer-Lambert law, while capturing four directional components for specular and body light in both forward and backward directions. Key contributions include on-the-fly geometry realization for yarns, implicit ply and fiber representations using elliptical cross-sections and 1D texture maps, a self-shadowing mechanism, and differentiable parameter fitting in Mitsuba 3, resulting in 3–5× speedups with reduced memory in near-field and ~20% faster distant rendering. The approach preserves fiber-level appearance, demonstrates strong agreement with ply-based references across yarns, knits, and woven fabrics, and enables scalable, high-fidelity cloth rendering suitable for design, production, and entertainment pipelines.

Abstract

Rendering realistic cloth has always been a challenge due to its intricate structure. Cloth is made up of fibers, plies, and yarns, and previous curved-based models, while detailed, were computationally expensive and inflexible for large cloth. To address this, we propose a simplified approach. We introduce a geometric aggregation technique that reduces ray-tracing computation by using fewer curves, focusing only on yarn curves. Our model generates ply and fiber shapes implicitly, compensating for the lack of explicit geometry with a novel shadowing component. We also present a shading model that simplifies light interactions among fibers by categorizing them into four components, accurately capturing specular and scattered light in both forward and backward directions. To render large cloth efficiently, we propose a multi-scale solution based on pixel coverage. Our yarn shading model outperforms previous methods, achieving rendering speeds 3-5 times faster with less memory in near-field views. Additionally, our multi-scale solution offers a 20% speed boost for distant cloth observation.

Figures (15)

• Figure 1: In this scene we compare the rendering results of our BYSDF model (near-field and multi-scale) to the reference ply-based model for a knitted beanie, with error maps provided in the insets. Our model achieves very similar results to the reference, both in distant and close-up views, with accurate soft shadows and geometry of plies and fibers as can be seen in the close-ups, in top and bottom rows respectively. Our approach achieves these results while utilizing only 20% of the memory at 2.5 times the speed of the reference rendering for the same quality (noise-level). Our multi-scale result offers additional performance gain by leveraging a level-of-detail strategy as shown in the distant results. The scene was lit using two area lights and one constant environment lighting.
• Figure 2: Implicit tracing using an elliptical yarn cross-section (CS) instead of a circular CS. The ray first intersects the yarn surface and cuts it into an ellipse. Our iterative approach (discussed in Section \ref{['ssec_geom_plies']}) calculates the intersection of ply-helices with the elliptical plane and then returns the closest ply by employing 2D ray tracing with a different $\TextOrMath{$v$\xspace}{\mathrm{v}}_\mathrm{ply}$.
• Figure 3: Implicit plies using an elliptical cross-section (CS). Circular CSs cause stretching or shortening in highly curved yarns (first row), leading to tangents differing significantly from the reference. In a slightly oblique view at 60$^\circ$ (below), the circular approximation introduces inaccuracies, causing an offset in overall shape and tangents compared to the reference, as marked by red arrows.
• Figure 4: Schematic overview of our iterative approach to finding the ellipse-helix intersection. The ray intersects the yarn surface, forming an ellipse in the side view. Newton iteration is used to find the helix-ellipse intersection, determining the ply hit and ply center, obtaining azimuthal phase $\TextOrMath{$u$\xspace}{\mathrm{u}}_\mathrm{ply}$ and longitudinal length $\TextOrMath{$v$\xspace}{\mathrm{v}}_\mathrm{ply}$ which is used to add fiber texture from 1D texture maps.
• Figure 5: Fiber migration, showing fibers appearing and disappearing to take into irregularities present in a yarn.