Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Parallel controllable texture synthesis
Lefebvre S., Hoppe H. ACM Transactions on Graphics (TOG)24 (3):777-786,2005.Type:Article
Date Reviewed: Nov 10 2005

The goal of texture synthesis is to generate large textures from a small exemplar image so the resulting texture does not contain obvious repetitions. This is a very active area of research in the graphics community, and a number of important papers have come out in this area in the past few years. This is one of them. Lefebvre and Hoppe report a parallel texture synthesis approach that allows real-time, on-demand generation of any area of an infinite aperiodic texture. The approach can be implemented in the graphics processing unit (GPU), and allows for a number of intuitive user controls, including the amount of randomness (at different locations or at different levels of resolution), the ability to drag and drop texture features, and the accurate reproduction of near-regular features. It does not have the obvious limitations of other approaches, such as tile-based methods (which map precomputed tiles whose distinctive features often reveal the tiling) or patch optimization (which iteratively uses precomputed areas of exemplar overlap to create a texture in an apparently inherently sequential process).

Using the pixel order-independent synthesis of Wei and Levoy [1], the authors describe an extension to the hierarchical synthesis Gaussian pyramid approach they call the Gaussian stack, which captures Gaussian pyramids shifted to all locations in the exemplar image. This allows for the jitter of upsampled coordinates in any level of the stack, creating spatially determined randomness at a desired level of detail in the resulting textures.

The paper is well written, with many more implementation details than are mentioned here. Many sample images are presented that nicely illustrate the features of the authors’ approach; empirical testing results are also provided. There are some very minor editing issues with the paper; for instance an accompanying video is mentioned (which is not referenced), and there are a couple of mistaken references to figures. Overall, this is a very nice paper, which I highly recommend to those in the graphics community.

Reviewer:  Myles F. McNally III Review #: CR132013 (0606-0649)
1) Wei,L. Levoy,M. Order-independent texture synthesis http://graphics.stanford.edu/papers/texture-synthesis-sig03/ (11/09/2005).
Bookmark and Share
  Featured Reviewer  
 
Texture (I.4.7 ... )
 
 
Lists, Stacks, And Queues (E.1 ... )
 
 
Parallel Algorithms (G.1.0 ... )
 
 
General (G.1.0 )
 
 
Data Structures (E.1 )
 
Would you recommend this review?
yes
no
Other reviews under "Texture": Date
Shape from texture: general principle
Kanatani K., Chou T. Artificial Intelligence 38(1): 1-48, 1989. Type: Article
Jan 1 1990
Analysis of fruit and vegetable surface color
Baranyai L., Szepes A. Machine Graphics & Vision International Journal 11(2/3): 351-361, 2002. Type: Article
May 29 2003
Texture and shape information fusion for facial action unit recognition
Kotsia I., Zafeiriou S., Nikolaidis N., Pitas I.  Advances in Computer-Human Interaction (Proceedings of the First International Conference on Advances in Computer-Human Interaction, Feb 10-15, 2008)77-82, 2008. Type: Proceedings
Sep 17 2008
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy