Achieve the look of Ukiyo-e for 3D animation

This research explores how the visual language of traditional Japanese ukiyo-e can be translated into a fully three-dimensional animation workflow. Ukiyo-e prints are known for their flat colour planes, strong contour lines, compressed spatial relationships, and material traces left by the woodblock printing process. While these qualities are often referenced in contemporary animation and visual effects, they are usually borrowed as isolated motifs rather than rebuilt as a complete visual system inside a 3D pipeline. In that sense, the main question of this project is not simply how to make a CG image look “Japanese” or “stylised,” but how to reconstruct a specifically ukiyo-e-like way of organising outlines, depth, texture, and camera design inside animated three-dimensional space. The project was conducted through practice-based research during the production of a short animated film in SideFX Houdini. Instead of separating technical development from creative production, the film itself functioned as the research environment. A series of iterative experiments was carried out using Solaris for rendering and COPs for image-based stylisation and compositing. These experiments focused on how selected visual characteristics of ukiyo-e could be adapted into a working NPR pipeline for animated shots rather than still images alone. Across the production process, three recurring constraints became clear. The first was a pixel constraint, where image-space outlines were limited by screen resolution and sampling behaviour. The second was a depth constraint, where flattening the image too aggressively removed useful spatial information and made scenes harder to read. The third was a camera constraint, where unrestricted camera motion and wider lenses reintroduced volumetric cues that conflicted with the intended planar aesthetic. These three constraints became the main analytical structure of the thesis. The final outcome of the research is a stylised rendering workflow that separates physically based rendering from post-process stylisation. Character and environment renders are first broken into AOVs, then reprocessed in COPs using cryptomatte-based selection, contour generation, position-based masking, wood grain overlays, and layered compositing. Rather than claiming a perfect recreation of historical printing, this research proposes a practical production framework for adapting ukiyo-e principles to contemporary 3D animation. The findings suggest that achieving this kind of stylised result depends not only on shader design, but also on how rendering, compositing, and cinematography are coordinated as one system.