“3D Model Partial-Resizing viaNormal and Texture Map Combination” by Nozawa, Fukusato and Morishima

  • ©Naoki Nozawa, Tsukasa Fukusato, and Shigeo Morishima

  • ©Naoki Nozawa, Tsukasa Fukusato, and Shigeo Morishima

  • ©Naoki Nozawa, Tsukasa Fukusato, and Shigeo Morishima

Conference:


Entry Number: 01

Title:

    3D Model Partial-Resizing viaNormal and Texture Map Combination

Presenter(s):



Abstract:


    Resizing of 3D model is necessary for computer graphics animation and application such as games and movies. In general, when users deform a target model, they built on a bounding box or a closed polygon mesh (cage) to enclose a target model. Then, the resizing is done by deforming the cage with target model. However, these approaches are not good for detailed adjustment of 3D shape because they do not preserve local information. In contrast, based on a local information (e.g., edge set and weight map), Sorkine et al. [Sorkine and Alexa 2007; Sorkine et al. 2004] can generate smooth and conformal deformation results with only a few control points. While these approaches are useful for some situations, the results depend on resolution and topology of the target model. In addition, these approaches do not consider texture (UV) information. To address these problems, we introduce a novel resizing method by combining geometry and texture information. Our system follows three steps: (i) generating geometry images, which store vertex position, normal and surface colors in similar 2D arrays using same implicit surface parameterization, from the target model; (ii) applying an image resizing (seam curving) into the images; and (iii) reconstructing the target model on the image space. In our system, we can resize the target model based on the image resolution. In addition, in order to prioritize the geometry of texture preservation for a specific region, our system allows the users to adjust the weights on 2D space.

Keyword(s):



Additional Images:

©Naoki Nozawa, Tsukasa Fukusato, and Shigeo Morishima ©Naoki Nozawa, Tsukasa Fukusato, and Shigeo Morishima

Acknowledgements:


    This work was supported by JSTACCEL Grant Number JPMJAC1602, Japan.


PDF:



ACM Digital Library Publication:



Overview Page:


Type: