Shadow Art Kanji: Inverse Rendering Application

📅 2025-03-15
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This work addresses the problem of generating high-fidelity, calligraphy-aesthetic Japanese Kanji shadows via inverse rendering. The method reconstructs a 3D geometric model—represented as a differentiable signed distance field (SDF)—that, under directional lighting, casts shadows matching target Kanji glyphs with precise structural and stylistic fidelity. Crucially, it introduces semantic constraints derived from Kanji character semantics as hard priors in the optimization objective, enabling artistic intent-driven reconstruction. The framework integrates differentiable rendering with a multi-view shadow consistency loss to ensure geometric plausibility and shadow alignment across viewpoints. Experiments on over 100 common Kanji characters achieve an average intersection-over-union (IoU) of >0.87, demonstrating accurate shadow topology and natural stroke rhythm. Reconstruction speed is five times faster than conventional approaches. This work establishes a novel paradigm for semantics-aware, illumination-driven artistic content generation.

Technology Category

Application Category

📝 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.
Problem

Research questions and friction points this paper is trying to address.

Balance artistic beauty and machine-generated imagery
Create 3D models casting Kanji-like shadows
Combine artistic expression with computational techniques
Innovation

Methods, ideas, or system contributions that make the work stand out.

3D models casting Kanji shadows
Combining art with computational techniques
Inverse rendering for shadow visualization
W
William Louis Rothman
College of Computing, Data Science, and Society, University of California, Berkeley
Yasuyuki Matsushita
Yasuyuki Matsushita
Microsoft Research Asia Tokyo
Embodied AIComputer Vision