Transcript
Page 1: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast and High Quality Overlap Fast and High Quality Overlap Repair for Patch-Based Texture Repair for Patch-Based Texture

SynthesisSynthesis

Andrew NealenAndrew NealenMarc AlexaMarc Alexa

Discrete Geometric Modeling Group (DGM)Discrete Geometric Modeling Group (DGM)Technische Universität DarmstadtTechnische Universität Darmstadt

Page 2: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Our Setting: 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.

Page 3: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Patch-Based Texture Synthesis

Some Existing Methods

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Page 4: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

Page 5: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

A B

Page 6: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

A B

Page 7: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

A B

Page 8: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

__22

A B

Page 9: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

__ ==22

overlap error

A B

Page 10: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

__ ==22

overlap error

A B

Page 11: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

__ ==22

overlap error

A B

Page 12: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

__ ==22

overlap error

A B

Page 13: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

Page 14: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► A Very Popular 2D Texture Synthesis Method• Image Quilting [Efros and Freeman 2001]• Graphcut Texures [Kwatra et. al 2003]• Wang Tiles [Cohen et. al 2003]

Patch-Based Texture Synthesis

Some Existing Methods

Page 15: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► Introduced at EGSR 2003 [Nealen and Alexa]

• Adaptive Patch Sampling, like Hierarchical Pattern Mapping [Soler et. al 2002]

• Per-Pixel Overlap Re-synthesis

Patch-Based Texture Synthesis

Hybrid Texture Synthesis (HTS)

Page 16: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Hybrid Texture SynthesisMethod

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Goal:From nxm, synthesize

NxM

similar, but not identical

Page 17: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Result (N x M)Result (N x M)

Goal:From nxm, synthesize

NxM

similar, but not identical

Input (n x m)

Intermediate Result

Hybrid Texture SynthesisMethod

Page 18: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Goal:From nxm, synthesize

NxM

similar, but not identical

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Patch-Search in the Input + Copy to Result + Mark Invalid Pixels

Hybrid Texture SynthesisMethod

Page 19: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Goal:From nxm, synthesize

NxM

similar, but not identical

Patch-Search in the Input + Copy to Result + Mark Invalid Pixels

Per-Pixel Re-synthesis Steps (for each Patch)

Hybrid Texture SynthesisMethod

Page 20: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Goal:From nxm, synthesize

NxM

similar, but not identical

Patch-Search in the Input + Copy to Result + Mark Invalid Pixels

Per-Pixel Re-synthesis Steps (for each Patch)

Hybrid Texture SynthesisMethod

Page 21: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Goal:From nxm, synthesize

NxM

similar, but not identical

Patch-Search in the Input + Copy to Result + Mark Invalid Pixels

Per-Pixel Re-synthesis Steps (for each Patch)

Hybrid Texture SynthesisMethod

Page 22: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Goal:From nxm, synthesize

NxM

similar, but not identical

Patch-Search in the Input + Copy to Result + Mark Invalid Pixels

Per-Pixel Re-synthesis Steps (for each Patch)

Hybrid Texture SynthesisMethod

Page 23: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Goal:From nxm, synthesize

NxM

similar, but not identical

Result (N x M)

Input (n x m)

Intermediate Result

Result (N x M)

Patch-Search in the Input + Copy to Result + Mark Invalid Pixels

Per-Pixel Re-synthesis Steps (for each Patch)

Hybrid Texture SynthesisMethod

Page 24: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Hybrid Texture SynthesisGeneralization: Pro and Con

► Pro: General Method for Overlap Repair• Complementary to other Methods, such as

Minimum-Error-Boundary-Cut (MEBC) or Feathering, yet oftentimes produces better results

• Generalizes to arbitrary patch shapes, i.e. is applicable to Graphcut Textures, Wang Tiles, etc.

► Con: Computationally Expensive• Exhaustive search for each invalid pixel in the

overlap, based on mostly irregular valid neighborhood

• Has O(rN log N) complexity -> Doesn‘t scale well.

Page 25: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Hybrid Texture SynthesisGeneralization: Pro and Con

► Pro: General Method for Overlap Repair• Complementary to other Methods, such as

Minimum-Error-Boundary-Cut (MEBC) or Feathering, yet oftentimes produces better results

• Generalizes to arbitrary patch shapes, i.e. is applicable to Graphcut Textures, Wang Tiles, etc.

► Con: Computationally Expensive• Exhaustive search for each invalid pixel in the

overlap, based on mostly irregular valid neighborhood

• Has O(rN log N) complexity -> Doesn‘t scale well.

Page 26: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Hybrid Texture SynthesisGeneralization: Pro and Con

► Pro: General Method for Overlap Repair• Complementary to other Methods, such as

Minimum-Error-Boundary-Cut (MEBC) or Feathering, yet oftentimes produces better results

• Generalizes to arbitrary patch shapes, i.e. is applicable to Graphcut Textures, Wang Tiles, etc.

► Con: Computationally Expensive• Exhaustive search for each invalid pixel in the

overlap, based on mostly irregular valid neighborhood

• Has O(rN log N) complexity -> Doesn‘t scale well.

Page 27: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap RepairBasic Idea

► Inspiration• Ashikhmin: Synthesizing Natural Textures

[2001] termed Coherence Search• Tong et. al‘s extension: k-Coherence Search

[2002]

► Basic Idea: Intelligently Reduce Search Space• Only search within a set of coherent pixels• Introduce Trade-off between quality and speed

Page 28: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

► Applying Coherence Search• For each pixel in the output, store its location in

the input in a source map (same size as the output texture)

Input Texture

Intermediate Result + Source Map

Fast Overlap Repair Coherence Search

Page 29: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair Coherence Search

► Applying Coherence Search• When searching for a new pixel, only consider

input pixels which are coherent with neighboring output pixels

Input Texture

Source Map Lookup

Intermediate Result + Source Map

Page 30: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair Coherence Search

► Applying Coherence Search• When searching for a new pixel, only consider

input pixels which are coherent with neighboring output pixels

Input Texture

Intermediate Result + Source Map

Source Map Lookup

Consequence in this example: Only two possible candidates

Page 31: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair Coherence Search

► Applying Coherence Search• Simply comparing to the coherent pixels results in

seams similar to Image Quilting (MEBC)

Example:

64x64 Texture Synthesized from four 32x32 Patches Coherence Exhaustive

Page 32: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair Coherence Search

► Applying Coherence Search• Simply comparing to the coherent pixels results in

seams similar to Image Quilting (MEBC)

Example:

64x64 Texture Synthesized from four 32x32 Patches Coherence Exhaustive

Page 33: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair Coherence Search

► Applying Coherence Search• Simply comparing to the coherent pixels results in

seams similar to Image Quilting (MEBC)

Example:

64x64 Texture Synthesized from four 32x32 Patches Coherence Exhaustive

Page 34: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair k-Coherence Search

Input Texture

Intermediate Result + Source Map

Source Map Lookup

► Better: Applying k-Coherence Search

Page 35: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair k-Coherence Search

► Better: Applying k-Coherence Search• Extend the set by the k-nearest neighbors (knn) of

each coherent pixel (in feature space) and remove duplicates

Intermediate Result + Source Map

Source Map Lookup

Input Texture

Page 36: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair k-Coherence Search

► Precomputation of knn Data Structure• Performed once for each nxm input texture and

stored for repeated use

• User defines size of box-shaped neighborhood np

• For each of the nxm input pixels─ Construct feature vector by ordered concatenation of the npx np

RGB-triples in the box-shaped neighborhood

• Dimension reduction (75-90%) by applying PCA• Compute indices of k-nearest neighbors to each

pixel

Page 37: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Fast Overlap Repair k-Coherence Search

► Source Map Maintenance• Each valid pixel in the overlap region is a linear

blend (feathering) of at least two original pixel values, i.e. from at least two different sources

• To avoid the maintenance of multiple source maps, simply store the source of the pixel with greatest contribution in a single source map

Blue: invalid overlap pixels

Page 38: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results varying k

k = 1

k = 11

k = 4

Exhaustive

Page 39: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results varying k

k = 1

k = 11

k = 4

Exhaustive

Page 40: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results varying k

k = 1

k = 11

k = 4

Exhaustive

Page 41: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results varying k

k = 1

k = 11

k = 4

Exhaustive

Page 42: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 43: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 44: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 45: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 46: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 47: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 48: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 49: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 50: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 51: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results timings

InputExhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

scales 64×64 δmax = 0.02 Δmax = 0.05

rock 128×128 δmax = 0.02 Δmax = 0.05

stonewall 200×200

δmax = 0.02 Δmax = 0.03

Pre: 0 sec.

Synth: 283 sec.

Pre: 0 sec.

Synth: 533 sec.

Pre: 0 sec.

Synth: 985 sec.

Pre: 6+3 sec.

Synth: 226 sec.

Pre: 45+62 sec.

Synth: 226 sec.

Pre: 247+28 s

Synth: 178 sec.

Pre: 6+4 sec.

Synth: 427 sec.

Pre: 45+74 sec.

Synth: 415 sec.

Pre: 247+37 s

Synth: 350 sec.

Page 52: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results

Exhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

Page 53: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Results

Exhaustive

n = 7x7

k-Coherence

n = 3x3 | k = 5

k-Coherence

n = 5x5 | k = 11

Page 54: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 55: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 56: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 57: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 58: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 59: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 60: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

ResultsSynthesis Comparisons

Input

Efros/Leung Wei/Levoy

IQ PBS HTS

Page 61: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Conclusions and Future Work

► Improve Error Metric• Still using the L2 norm due to its simplicity

• Develop a metric which takes feature mismatch into account

• Texton map approach [Zhang et al. 2003]• Feature Map [Wu and Yu 2004] performs even

better, and for near-regular textures, see [Liu et. al 2004] (both to appear at SIGGRAPH 2004)

Page 62: Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004 Fast and High Quality Overlap Repair for Patch-Based Texture Synthesis

Andrew Nealen and Marc Alexa, Discrete Geometric Modeling Group, TU Darmstadt, 2004

Questions ?

► Contact Information

Andrew [email protected]

Marc [email protected]

http://www.dgm.informatik.tu-darmstadt.de

Matlab code:http://www.dgm.informatik.tu-darmstadt.de/research/texsynth.html


Top Related