2.5D cartoon modelling is a recently proposed technique for modelling 2D cartoons in 3D, and enables 2D cartoons to be rotated and viewed in 3D. Automatic modelling is essential to efficiently create 2.5D cartoon models. Previous approaches to 2.5D modelling are based on manual 2D drawings by artists, which are inefficient and labour intensive. We recently proposed an automatic framework, known as Automatic 2.5D Cartoon Modelling (Auto-2CM). When building 2.5D models using Auto-2CM, the performance of different algorithm configurations on different kinds of objects may vary in different applications. The aim of perceptual evaluation is to investigate algorithm selection, i.e. selecting algorithm components for specific objects to improve the performance of Auto-2CM. This paper presents experimental results on different algorithms and recommends best practice for Auto-2CM.
展开▼
