Blobtree modelling
暂无分享,去创建一个
A Blobtree is a 3D scalar field which can represent volumes or surfaces. The research in this thesis describes a framework for Blobtree modelling. The contributions of the research concern three crucial parts of the modelling system.
First of all, fast visualization is required to allow for a responsive modelling system. The RaySkip method exploits object space and temporal coherency by estimating new surface point positions based on previously found surface points. This improves the rendering times of high quality previews and the final animation/image. More importantly, the Turbo Charged Blobtree drastically improves visualization times in general by using space subdivision and a new Blobtree data structure. Space subdivision requires the creation of many copies of the original Blobtree model. The new data structure reduces the construction times of each copy and will also reduce memory requirements.
The second part of the modelling system concerns shape modelling. Blending between components can be achieved by summating their scalar fields. Whereas existing blend operations offer limited or no control of the resulting shape, the Locally Restricted Blend operation allows the user to manipulate the blend shape between each pair of components. This is achieved by deforming the scalar field of each of the blend components. Besides a new blend operation a new deformation operation is introduced which allows a variety of deformations to be applied to Blobtree models. This deformation operation can deform the scalar field of a Blobtree model by applying a transformation weighted by a second Blobtree model. Using the Turbo Charged Blobtree, the new blend and deformation operations can be applied interactively.
The third and final part of thee modelling system in which this research makes contributions is texturing. There are many texturing methods which contribute towards a wide variety in texture types, but existing methods do not apply to Blobtree models or have other limitations. The new Implicit Decal method allows the user to automatically apply cellular patterns to any model representation including Blobtrees. This is done by covering the surface of the model with spherical implicit fields. The fields are deformed based on each others proximity to fit a Voronoi diagram instead of overlapping each other, which creates a cellular pattern. To project other types of images onto a model, the Implicit Texture Coordinate Interpolation method can be used to create 2D and 3D textures for arbitrary shapes. This is done by placing a set of spherical fields on or around the surface of the model. By using interpolation, a 2D parameterization is calculated which can be used for texturing.