Shadow Art Kanji: Inverse Rendering Application

TL;DR

This work addresses rendering Kanji-like shadows by reframing inverse rendering as a voxel-occupancy problem. It first develops a linear programming formulation that relaxes binary voxels to continuous values and derives projection-based constraints from three orthogonal planes, then circumvents slow optimization with a direct carving algorithm that iteratively removes voxels along plane normals. Empirical results show the direct carving approach yields higher-resolution, visually appealing shadows with substantially faster runtimes than the LP, and the method scales to higher dimensions and printable outputs. The work demonstrates a practical pipeline from Kanji input to 3D-printed shadow-casting objects, highlighting a significant improvement in efficiency and printability while generalizing to multi-planar and higher-dimensional projections.

Abstract

Finding a balance between artistic beauty and machine-generated imagery is always a difficult task. This project seeks to create 3D models that, when illuminated, cast shadows resembling Kanji characters. It aims to combine artistic expression with computational techniques, providing an accurate and efficient approach to visualizing these Japanese characters through shadows.

Figures (10)

• Figure 1: Mitsuba's render and shadow b1.
• Figure 2: Graph showing the relationship between the global threshold λ and the similarity percentage (the fraction of projection pixels matching the target projection) when $\epsilon = 0.239$. The graph indicates a sharp increase in similarity at $\lambda \approx 0.239$. Indeed, additional tests indicate that an optimal threshold value of $\lambda = \underset{\delta \to 0^+}{\lim} \delta + \epsilon$.
• Figure 3: This rendered mesh demonstrates the current model's aliasing problem.
• Figure 4: Traditional anti-aliasing by supersampling b6.
• Figure 5: A rendered, anti-aliased mesh with $(n, N) = (32, 8)$.