Data-driven Architectural texture mapping - PowerPoint PPT Presentation

data driven architectural texture mapping n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Data-driven Architectural texture mapping PowerPoint Presentation
Download Presentation
Data-driven Architectural texture mapping

play fullscreen
1 / 19
Data-driven Architectural texture mapping
141 Views
Download Presentation
lenora
Download Presentation

Data-driven Architectural texture mapping

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. 由于建筑物的3D model和textures均属于structured data,故是否题目扩展为structural? Data-driven Architectural texture mapping Textured Architectures Texture mapping Un-textured 3D scene Textured output

  2. 1. Motivations • Simplify the tedious/challenging texture mapping • Data: architecture 3D models + structured textures Textured Architectures Un-textured 3D scene Texture mapping Textured output

  3. Textured architectural model properties • Data properites: • Simple geometric surfaces (polygons) • Attractive textures (always texel)

  4. Architectural textures? • Textures for buildings, interiors • Highly structured content

  5. Difficulties • Large, interactive environments • Correlation: surfaces  textures • Small set of carefully designed textures • Used on a variety of surfaces ?

  6. different texturesvsdifferent surfaces • Surfaces-textures correspondence • Part/region level • Neighboring propagation • Textures and surfaces • Tiling + cropping • Cut features • Scaling • Distortions ? *****

  7. 2. Previous work Color Transfer Reinhard et al: Color Transfer between Images, IEEE CG&A 2001 Analogy Hertzmann et al: Image Analogies, Siggraph 2001 Mertens et al: Texture Transfer Using Geometry Correlation, EGSR 2007 Colorization Welsh et al: Transfering Color to Greyscale Images, Siggraph2001 Irony et al: Colorization by Examples, EGSR 2005

  8. Texture synthesis, retargetting • By example texture synthesis [Wei and Levoy 2001, Efros and Freeman 2001, …] • On the fly [Lefebvre and Hoppe 2005] • Limited to stochastic textures • Retargetting, summarization, reshuffling [Avidan and Shamir 2007, Simakov et al. 2008,…] • High quality • Pre-computed: One image per surface

  9. Specialized to content • Grammar-based approaches [Mueller et al. 2007, Teboul et al. 2010 … ] • Great for facades • In pixel shader[Haegler et al. 2010 ] • Requires content description • Proceduraltextureapproaches [Lasramet al. 2012… ]

  10. 3D Material style transfer • 3D Material style transfer [Nguyen et al. 2012 EG] • Two steps: • material extraction • material assignment extracts materials from a 2D guide source and assigns it to a resulting target 3D scene.

  11. (Procedural) texture • Feature-based Cellular Texturing for Architectural Models [Legakis et al. 2001, siggraph] • Cellular textures→basemesh • Aligning to edges, wrapping around corners, responding to annotations supplied by the designers

  12. Procedural Example synthesis • By-example Synthesis of Architectural Textures [Lefebvre et al. 2010, siggraph] • Synthesis of structured textures • User control • Decoded at render-time

  13. Texture Transfer Using Geometry Correlation • [Mertens et al. 2006, Euro Symposium on Rendering] • local geometric features • Texture discriminations: • curvature + orientation Source: mesh + texture input Texture Transfer Output with texture Target: mesh

  14. Assisted texture assignment • [Chajdaset al. 2010, Si3D] **** • Surface similarity • Scoring texture candidate with two factors: • Surface similarity + neighboring consistency (a) Initial scene. (b) The user selects a texture for a floor (hilighted in yellow), all floors are automatically assigned by our system. (c) The user selects a texture for a border, all borders are automatically assigned. (d) Result after 10 selections. Surfaces touched by the user are highlighted in yellow. Texturing information is propagated throughout the entire scene at once.

  15. Assisted texture assignment • Architectural surface probes • These abstract features are specifically designed for matching surfaces in a texturing context Example scene, local orientation (normals), anisotropy, curvature, distance to edge, and accessibility.

  16. 3. Some insights • The surface similarity [Chajdas et al. 2010] • abstracts the actual shape of the polygon while retaining information relevant for texturing. • The textures scoring [Chajdas et al. 2010] • Ranking each texture according to whether it is a likely candidate for a given surface. • Example synthesis [Lefebvre et al. 2010] • Architectural texture property • Optimization [Nguyen et al. 2012] • the prohibitively large search space

  17. Example synthesis Insights • Properties of architectural textures • Auto-similar by translation • Parallel path (cuts) having similar colors left,right

  18. Optimization – Simulated Annealing ϴ ϴ ϴ ϴ + + ϴ ϴ ϴ ϴ assignment i+1th ϴ ϴ ϴ ϴ otherwise … … … … assignment ith

  19. Cheers Happy