Texture Synthesis

Texture Synthesis

No Texture

Using a sampling of 2D perlin Noise provides smooth hills.

Terrain Coloring Using a 1D texture map based on the

altitude can provide many useful mapping.

Terrain Coloring

Striped 1D texture map.

Terrain Coloring

Using a 2D texture map provides richer detail, but is independent of the terrain.

Terrain Coloring More advanced coloring is based on

altitude and slope.

Rolling Hills

Flowers, trees, paths, …

Procedural texture synthesis [Gardner 85] sum a small number of sine

waves of different periods and amplitudes.

[Perlin 85]use band-limited noise function [Turing 52] Reaction-diffusion system in

developmental biology.

Two or more chemicals can diffuse through an embryo and react with each other until a stable pattern of chemical concentration is reached.

Reaction-diffusion

Texture buttons – Andrew Witkin and Michael Kass

Texture synthesis on surfaces

Retiling– Sprinkle some new vertices onto the mesh– Space them uniformly over the mesh using point

repulsion.– Remove old vertices and re-triangulate– Create a hierarchy of meshes

Neighborhood Construction: – flatten a set of nearby vertices– resample the flattened patch

Point Repulsion

Point Repulsion

Mutual Tesselation

Neighborhood Construction

Neighborhood Construction

Texture Synthesis on meshes

Texture Synthesis on meshes

Texture Synthesis on meshes

2D Texture Synthesis

nxm Input Texture

NxM Output Texture The goal: Synthesize an output texture

which is perceptually similar to the input texture. Also ensure that the result contains sufficient variation.

Tiled Textures

Take up little texture memory

Visual Comparison

Synthesized textureSimple tiling

Synthetic tilabletexture

Existing (and Impressive) Technology

Pixel-Based:– Non-parametric Sampling [Efros and Leung 1999]– Tree-structured Vector Quantization [Wei and

Levoy 2000]– Synthesizing Natural Textures [Ashikhmin 2001]– Image Analogies [Hertzmann et al. 2001]

Patch-Based:– Patch-Based Sampling [Guo et al. 2001]– Image Quilting [Efros and Freeman 2001]– Hierarchical Pattern Mapping [Soler et al. 2002]

Pixel-Based Texture Synthesis

Input Output

Step n in Algorithm

Pixel-Based Texture Synthesis

Input Output

Build L-shaped Neighborhood

Pixel-Based Texture Synthesis

Input Output

Find Matching Neighborhood(s) in Input

Pixel-Based Texture Synthesis

Input Output

Select Best Neighborhood from all Candidates

Pixel-Based Texture Synthesis

Input Output

Copy Selected Pixel to Output

Pixel-Based Texture Synthesis

Input Output

Repeat for all Pixels in this Row...

Pixel-Based Texture Synthesis

Input Output

...and all other Output Rows

Patch-Based Texture Synthesis

Input Output

Step n in Algorithm

Patch-Based Texture Synthesis

Input Output

Patch to be Synthesized in this Step

Patch-Based Texture Synthesis

Input Output

Build Neighborhood (Image Mask)

Patch-Based Texture Synthesis

Input Output

Find Matching Neighborhood(s) in Input

Patch-Based Texture Synthesis

Input Output

Select Best Neighborhood from all Candidates

Patch-Based Texture Synthesis

Input Output

Copy Selected Patch to Output

Patch-Based Texture Synthesis

Input Output

Apply some Overlap Repair Strategy ...

Patch-Based Texture Synthesis

Input Output

... And we are done with this patch

Patch-Based Texture Synthesis

Input Output

Repeat for all Patches in Output

Texture Synthesis

Solid texture from 2D views– Heeger & Bergen 1995– Ghazanfarpour & Dischler 1999

Images from [Heeger&Bergen 1995]

2D Source 3D result

Previous Work

Texture varying, morphing, and mixture– Portilla & Simoncelli 1999– Bar-Joseph, El-Yaniv, Lichinski, Werman 2001– Z. Liu, C. Liu, Shum, Yu 2003– PVT paper in SIGGRAPH 2003

Image from [Portilla&Simoncelli 1999]

Source 1 Source 2morphing

Brodatz Resultsaluminum wirereptile skin

More Brodatz Results

french canvas rafia weave

More Resultswood granite

More Resultswhite bread brick wall