Publications

Patternista: Learning Compatible Elements and Composition for Plate Decoration

International Symposium on Non-Photorealistic Animation and Rendering (NPAR)

Publication date: May 1, 2016

Quoc Huy Phan, Paul Asente, Jingwan (Cynthia) Lu, Hongbo Fu, Antoni Chan

Creating aesthetically pleasing decorations for daily objects is a task that requires the designer to have a deep understanding on multiple aspects of object decoration, including color, composition and element compatibility. Moreover, a designer also needs a dedicated sense of style to create works that stand out from the crowd. Although researchers have discretely carried out researches on some of the aspects of object decoration [GAGH14], the overall problem of object decoration is still considered untouched. In this chapter, we propose a flexible data-driven framework to jointly consider many aspects of the problem. We developed a ring-based layout that is capable of capturing decorative compositions not only for plates but also other objects such as vases and pots, which we call “ring-based objects”. Our representation of the layout allows us to flexibly infer the properties of each decorative region in a sequential fashion. Our learning model, which is based on Hidden Markov Model, is capable of generating arbitrarily complex decorations without sacrificing the performance. We conducted both quantitative and qualitative experiments to evaluate the framework and obtained very favorable results.

Learn More