Creating Textures for Realistic Images Multiresolution Textures from Image Sequences Extracting Textures From

endering is one of the most important tasks in computer graphics and animation , and texture maps add essential visual content to the rendered image. Unfortunately, extracting textures from a single photograph poses severe difficulties and is sometimes impossible, while artificial texture synthesis does not address the full range of desired textures. We present a method for computing high-quality, multiresolution textures from an image sequence. Our technique has the following features: 1. It can be used with images in which the textures are present in different resolutions and different perspective distortions. 2. It can extract textures from objects with any known 3D geometric structure; specifically, we are not restricted to planar textures. 3. It removes directional illumination artifacts such as highlights and reflections. 4. It stores the resulting texture efficiently in a multi-resolution data structure. 5. It imposes no restrictions on the computed texture, which can be a constant color or richly colored. One particularly attractive application of our method, illustrated in the conclusion, produces animation sequences of existing objects endowed with synthetic behavior. This emphasizes the advantage of the multiresolution representation, since each frame uses the level of detail it needs for any location in the texture. In addition to the geometry of the rendered object, high-quality rendering requires the object's material properties, texture maps associated with the object's surfaces, and algorithms to produce images given these data. The simulation of light is well understood. 1 Artificial texture generation, on the other hand, still involves some unsolved difficulties. Although specific textures have been simulated successfully, artificially generated textures usually look too " clean, " even when noise is added using statistical methods. In practice , you must use textures from real images (such as photographs or video stills) or textures created using 2D paint systems to create satisfactory visual results. Certain applications inherently require textures from real images. For example, to endow a real object with a personality and orchestrate its motion with animation software, we would naturally want its real textures to be available to the rendering system. But computing texture maps from a real image poses several difficulties: s The lighting conditions under which the image was taken might not match the desired illumination on the texture-mapped object. Specifically, the image usually contains specular light effects such as highlights and reflections that are not supposed to be seen when the texture is mapped onto an object. s …

[1]  Peter J. Burt,et al.  Enhanced image capture through fusion , 1993, 1993 (4th) International Conference on Computer Vision.

[2]  Lawrence B. Wolff,et al.  Scene understanding from propagation and consistency of polarization-based constraints , 1994, 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[3]  Steven A. Shafer,et al.  Using color to separate reflection components , 1985 .

[4]  James T. Kajiya,et al.  The rendering equation , 1986, SIGGRAPH.

[5]  Ari Rappoport,et al.  Three‐Dimensional Modeling and Effects on Still Images , 1996, Comput. Graph. Forum.

[6]  A Blake,et al.  Shape from specularities: computation and psychophysics. , 1991, Philosophical transactions of the Royal Society of London. Series B, Biological sciences.

[7]  Lance Williams,et al.  Pyramidal parametrics , 1983, SIGGRAPH.

[8]  David Salesin,et al.  Multiresolution painting and compositing , 1994, SIGGRAPH.

[9]  Sang Wook Lee,et al.  Detection of Specularity Using Color and Multiple Views , 1992, ECCV.

[10]  Michal Irani,et al.  Super resolution from image sequences , 1990, [1990] Proceedings. 10th International Conference on Pattern Recognition.