.png){kind=icon}
.png){kind=icon}
Return
3D shape creation by style transfer
.png){kind=icon}
PRE
AI
DOI：10.1007/s00371-014-0999-1.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In this paper, we propose a new style transfer method for automatic 3D shape creation based on new concepts of style and content of 3D shapes. Our unsupervised style transfer method could plausibly create novel shapes not only by recombining existent styles and contents in a set but also by combining new-coming styles or contents with the existent ones conveniently. This feature provides a better way to increase the diversity of created shapes. The process of shape creation can be summarized as two stages. First, style and content separation is performed to analyzed shapes in a set. Second, novel shapes are created by style transfer. In our setting, contents are first separated via clustering shapes using a new defined global shape distance, and then, style parts are clustered into different style classes. Specifically, style parts are extracted from each pair of intra-content shapes through comparing their multi-scale corresponding patches instead of corresponding parts. This strategy makes the process of extracting style parts become insensitive to slight geometric changes. The multi-scale corresponding patches are obtained via partitioning the two shapes in a consistent way by the proposed correspondence transfer. Meanwhile, to quantify the comparison results for locating style parts, a novel local shape difference function (LSDF) is introduced. Based on LSDF, extracting a style part from each shape is formulated as an optimal LSDF threshold selection problem. In the experiments, we test our method in several sets of man-made 3D shapes and obtain plausible created shapes based on the reasonably separated styles and contents.
Keywords:
Style and content separation
Style transfer
Shape creation
Local shape difference function
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
2.9
Papers:
4.5K
Citations:
6.5K
Organization
View more organizations