EC project no. IST-2001-34744 RealReflect 
    MSMT grant 1M6798555601 DAR
 
Texture Tiling and Patching With a Novel Path Search Algorithm for Quick and Realistic Texture Synthesis
  Petr Somol     Michal Haindl
Abstract:
We present a framework for sampling-based texture synthesis where the learning phase is clearly separated and the synthesis phase is very simple and fast. The approach exploits the potential of tile-based texturing and produces good and realistic results for a wide range of textures. The synthesis phase is realizable directly in graphical hardware. As a part of the framework we also propose a fast and adjustable sub-optimal path search algorithm for finding minimum error boundaries between overlapping images. The algorithm may serve as an efficient alternative of traditional slow path search algorithms like the dynamical programming etc.

  • + Analysis (tile creation) deterministic or randomized
  • + Surprisingly good results with many natural textures
  • + Synthesis trivial: just combining rectangular tiles - can be easily done in GPU
  • + Adjustable trade-off between the required quality of output and memory size (number of different tiles)
  • + Includes extremely fast sub-optimal path-search algorithm used to minimize overlap errors between bitmaps
  • +/- Analysis may involve parameter twiddling
  • - Can fail e.g. with textures that exhibit strong linear diverging structures
    Demo (requires Flash version 5+ plug-in)
    Sample Results

      
      
      
      
      
      
      
      
      


    Further information can be found in:

    Somol, P. - Haindl, M.
    Novel Path Search Algorithm for Image Stitching and Advanced Texture Tiling
    In: 13th International Conference in Central Europe
    on Computer Graphics, Visualization and Computer Vision, WSCG'2005
    Plzen (Pilsen), Czech Republic
    V.Skala Ed., ISBN 80-903100-7-9,
    University of West Bohemia, pp. 155 - 162.
  • Conference slides can be downloaded here as PDF file (6MB).
  • Research Report containing slightly extended version of the text is available here (10MB). 

    • 


    

    

    




   Back to the demonstration page.