87
Model-Driven 3D Content Creation as Variation Hao (Richard) Zhang – 张张 GrUVi Lab, Simon Fraser University (SFU) Talk @ HKUST, 04/21/11 TAU ZJU NUDT SFU

Model-Driven 3D Content Creation as Variation

  • Upload
    berny

  • View
    41

  • Download
    0

Embed Size (px)

DESCRIPTION

NUDT. TAU. ZJU. SFU. Model-Driven 3D Content Creation as Variation. Hao (Richard) Zhang – 张皓 GrUVi Lab, Simon Fraser University (SFU) Talk @ HKUST, 04/21/11. 3D content creation. Inspiration?. Inspiration  a readily usable digital 3D model. Realistic reconstruction. - PowerPoint PPT Presentation

Citation preview

Page 1: Model-Driven 3D Content Creation as Variation

Model-Driven 3D Content Creation as

Variation

Model-Driven 3D Content Creation as

VariationHao (Richard) Zhang – 张皓

GrUVi Lab, Simon Fraser University (SFU)

Talk @ HKUST, 04/21/11

Hao (Richard) Zhang – 张皓GrUVi Lab, Simon Fraser University (SFU)

Talk @ HKUST, 04/21/11

TAUTAU ZJUZJUNUDTNUDTSFUSFU

Page 2: Model-Driven 3D Content Creation as Variation

3D content creation

Inspiration a readily usable digital 3D modelInspiration a readily usable digital 3D model

Inspiration?Inspiration?

Page 3: Model-Driven 3D Content Creation as Variation

Realistic reconstruction

• Inspiration = real-world data

[Nan et al., SIGGRAPH 2010][Nan et al., SIGGRAPH 2010]

Page 4: Model-Driven 3D Content Creation as Variation

Creative inspiration

•Creation of novel 3D shapes

• Inspiration = design concept, mental

picture, …

sketchsketch

High demand in VFX, games, simulation, VR, …

High demand in VFX, games, simulation, VR, …

Page 5: Model-Driven 3D Content Creation as Variation

3D content creation is hard

•2D-to-3D: an ill-posed problem

▫Shape from shading, sketch-based modeling, …

•Creation from scratch is hard: job for skilled

artistsOne of the most central problems in

graphics; One of the most discussed at SIG’10 panel

One of the most central problems in graphics; One of the most discussed at

SIG’10 panel

Page 6: Model-Driven 3D Content Creation as Variation

Usable 3D content even harder

•Models created are meant for subsequent

use

•Creation of readily usable 3D models

Page 7: Model-Driven 3D Content Creation as Variation

Usable 3D content even harder

•Models created are meant for subsequent use

•Creation of readily usable 3D models

•Higher-level information beyond low-level mesh

▫Part or segmentation information

▫Structural relations between parts

▫Correspondence to relevant models, etc.Hard shape analysis problems!Hard shape analysis problems!

Page 8: Model-Driven 3D Content Creation as Variation

Key: model reuse

•Reuse existing 3D models and associated

info

•Model-driven approach: creation is driven

by or based on existing (pre-analyzed)

models

Page 9: Model-Driven 3D Content Creation as Variation

Key: model reuse

•Reuse existing 3D models and associated

info

•Model-driven approach: creation is driven

by or based on existing (pre-analyzed)

models

•Two primary modes of reuse:

▫New creation via part composition

Page 10: Model-Driven 3D Content Creation as Variation

Key: model reuse

•Reuse existing 3D models and associated info

•Model-driven approach: creation is driven by

or based on existing (pre-analyzed) models

•Two primary modes of reuse:

▫New creation via part composition

▫New creation as variation or modification of

existing model(s), e.g., a warp or a

deformation

Page 11: Model-Driven 3D Content Creation as Variation

Modeling by example

•New models composed by parts retrieved

from an existing data repository

•Key: retrieve relevant parts

•Many variants …

[Funkhouser et al., SIGGRAPH 2004][Funkhouser et al., SIGGRAPH 2004]

Page 12: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Significant deviation from existing models

▫Exploratory modeling via part suggestions

[Chaudhuri & Koltun., SIG Asia 2010][Chaudhuri & Koltun., SIG Asia 2010]

Page 13: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Significant deviation from existing models

▫Exploratory modeling with part suggestions

•Cons:▫Are models composed by parts readily

usable?

Page 14: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Significant deviation from existing models

▫Exploratory modeling with part suggestions

•Cons:▫Are models composed by parts readily

usable? structure lost by part composition; how to stitch?

Page 15: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Significant deviation from existing models

▫Exploratory modeling with part suggestions

•Cons:▫Are models composed by parts readily

usable? structure lost by part composition; how to stitch?

▫Does part exploration always reflect user design intent?

Page 16: Model-Driven 3D Content Creation as Variation

Model-driven creation as variation

•New creation as variation of existing model(s)

Enrich a set; generate “more of

the same” …

Enrich a set; generate “more of

the same” …

Photo-inspired 3D model creation

Photo-inspired 3D model creation

Inspiration = photographsInspiration = photographs

Inspiration = a model set

Inspiration = a model set

Page 17: Model-Driven 3D Content Creation as Variation

Model-driven creation as variation

•New creation as variation of existing model(s)

Enrich a set; generate “more of

the same” …

Enrich a set; generate “more of

the same” …

Inspiration = a model set

Inspiration = a model set

Page 18: Model-Driven 3D Content Creation as Variation

Style-Content Separation by Anisotropic Part Scales

Kai Xu1,2, Honghua Li2, Hao Zhang2, Daniel Cohen-Or3

Yueshan Xiong2, and Zhi-Quan Cheng2

1Simon Fraser Universtiy 2National Univ. of Defense Tech. 3Tel-Aviv University

Page 19: Model-Driven 3D Content Creation as Variation

Motivation

•Enrich a set of 3D models with their derivatives

Set belongs to the same family or class

Set belongs to the same family or class

Page 20: Model-Driven 3D Content Creation as Variation

Variations in shape parts in the set

Geometric or content difference

Part proportion (= style) difference

Page 21: Model-Driven 3D Content Creation as Variation

?

Style transfer as a derivative

Part proportion style

Page 22: Model-Driven 3D Content Creation as Variation

?

Style transfer as a derivative

Part proportion style

Page 23: Model-Driven 3D Content Creation as Variation

Difficulty with style transfer

•Style transfer needs part correspondence

•Part correspondence is difficult

▫Unsupervised problem

▫Both content and style variations

Variations can be significant!

Page 24: Model-Driven 3D Content Creation as Variation

Work at part and OBB level

Parts enclosed and characterized by tight oriented bounding boxes (OBBs)Parts enclosed and characterized by tight oriented bounding boxes (OBBs)

Page 25: Model-Driven 3D Content Creation as Variation

Style content separation

•To address both shape variations in the set▫Separate treatment of “style” and “content”

Style 1

Style 2

Style 3

ContentContent

Sty

leS

tyle

Page 26: Model-Driven 3D Content Creation as Variation

Style transfer as a derivative

•Creation = filling in the style-content table

Page 27: Model-Driven 3D Content Creation as Variation

Style vs. content

•Fundamental to human perception

Content Style

Language Words Accents

Text Letters Fonts

Human face Identities Expressions

Page 28: Model-Driven 3D Content Creation as Variation

Style content separation

•Previous works on faces, motion, etc.

▫Prerequisite: data correspondence

▫Correspondence dealt with independently

▫Correspondence itself is the very challenge!

Page 29: Model-Driven 3D Content Creation as Variation

Our approach

• One particular style:

Anisotropic part scales or part proportions

Page 30: Model-Driven 3D Content Creation as Variation

Our approach

• One particular style:

Anisotropic part scales or part proportions

• The approach:

Style-content separation with style

clustering in a correspondence-free way

Page 31: Model-Driven 3D Content Creation as Variation

Algorithm overview

•Pipeline

Style clustering Co-segmentation Inter-style part correspondence

Contentclassification

Page 32: Model-Driven 3D Content Creation as Variation

Anisotropic part scales

•Measure style distance between two shapes

Page 33: Model-Driven 3D Content Creation as Variation

Anisotropic part scales

•Measure style distance between two shapes

Part OBB graphs of

given segmentatio

n

Page 34: Model-Driven 3D Content Creation as Variation

Anisotropic part scales

•Measure style distance between two shapes

Computestyle

signatures

……

Part OBB graphs of

given segmentatio

n

Page 35: Model-Driven 3D Content Creation as Variation

Anisotropic part scales

•Measure style distance between two shapes

……

Part OBB graphs of

given segmentatio

n

Euclideandistance

Computestyle

signatures

Page 36: Model-Driven 3D Content Creation as Variation

Style distance issues

•Unknown segmentation

•Unknown correspondence

?

?

Page 37: Model-Driven 3D Content Creation as Variation

Style distance

•Search over all part compositions and part counts

……

……

Page 38: Model-Driven 3D Content Creation as Variation

Style distance

•For each part count, find minimal distance

……

……

A good signature will return min distance across all part counts to reflect corresponding part decompositions …

Page 39: Model-Driven 3D Content Creation as Variation

Correspondence-free style signature

Binary relations: difference of part scales between adjacent OBBs

Use Laplacian graph spectra:

OBB graph

Page 40: Model-Driven 3D Content Creation as Variation

Correspondence-free style signature

Unary attributes: anisotropy of parts

Use Laplacian graph spectra:

OBB graph linear planar spherical

Graph spectra is permutation-free

Page 41: Model-Driven 3D Content Creation as Variation

Style clustering

•Spectral clustering using style distances

Page 42: Model-Driven 3D Content Creation as Variation

Pipeline

Style clustering Co-segmentation Inter-style part correspondence

Contentclassification

Page 43: Model-Driven 3D Content Creation as Variation

Co-segmentation

•Approach:▫Consistent segmentation [Golovinskiy & Funkhouser,

SMI 09]

▫ Initial guess: global alignment (ICP)

[Golovinskiy & Funkhouser 09]

Page 44: Model-Driven 3D Content Creation as Variation

Co-segmentation

•Approach:▫Consistent segmentation [Golovinskiy & Funkhouser, SMI 09]

▫ Initial guess: global alignment (ICP)

•We co-segment within a style cluster▫Removing non-homogeneous part scaling from analysis

[Golovinskiy & Funkhouser 09]

Page 45: Model-Driven 3D Content Creation as Variation

Co-segmentation

•Approach:▫Consistent segmentation [Golovinskiy & Funkhouser, SMI 09]

▫ Initial guess: global alignment (ICP)

•We co-segment within a style cluster▫Removing non-homogeneous part scaling from analysis

[Golovinskiy & Funkhouser 09] After style separation

Page 46: Model-Driven 3D Content Creation as Variation

Pipeline

Style clustering Co-segmentation Inter-style part correspondence

Contentclassification

Page 47: Model-Driven 3D Content Creation as Variation

Inter-style part correspondence

•Approach: deform-to-fit

▫Deformation-driven correspondence [Zhang et al., SGP 08]

▫Consider common interactions between OBBs

1D-to-1D 1D-to-2D 2D-to-2D 2D-to-3D

Page 48: Model-Driven 3D Content Creation as Variation

Inter-style part correspondence

•Deform-to-fit: appropriate deformation energy

Pruned priority-driven search

Page 49: Model-Driven 3D Content Creation as Variation

Pipeline

Style clustering Co-segmentation Inter-style part correspondence

Contentclassification

Page 50: Model-Driven 3D Content Creation as Variation

Content classification

•Use Light Field Descriptor [Chen et al. 2003]

•Compare corresponding parts

Part-level LFD Global LFD

Page 51: Model-Driven 3D Content Creation as Variation

Synthesis by style transfer

•OBBs are scaled

•Underlying geometry via space

deformationcontent

style style transfer

Page 52: Model-Driven 3D Content Creation as Variation

Results: hammers

Page 53: Model-Driven 3D Content Creation as Variation

Results: goblets

Page 54: Model-Driven 3D Content Creation as Variation

Results: humanoids

Page 55: Model-Driven 3D Content Creation as Variation

Results: chairs

Page 56: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Automatic generation of many variations

▫Unsupervised

▫Deals with anisotropic part scales

▫Variation = part scaling: structure preservation

Page 57: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Automatic generation of many variations

▫Unsupervised

▫Deals with anisotropic part scales

▫Variation = part scaling: structure preservation

•Cons:▫Rely on sufficiently good initial segmentations

▫Variation does not create new content

Page 58: Model-Driven 3D Content Creation as Variation

Interesting future work

•Learn and synthesize with generic styles

Page 59: Model-Driven 3D Content Creation as Variation

Model-driven creation as variation

•New creation as variation of existing model(s)

Photo-inspired 3D model creation

Photo-inspired 3D model creation

Inspiration = photographsInspiration = photographs

Page 60: Model-Driven 3D Content Creation as Variation

Photo-inspired 3D modeling

Photo-Inspired Model-Driven 3D Object Modeling

Kai Xu1,2, Hanlin Zheng4, Hao Zhang2, Daniel Cohen-Or3

Ligang Liu4, and Yueshan Xiong2

1NUDT 2SFU 3TAU 4ZJU Conditionally acceptedConditionally accepted

Page 61: Model-Driven 3D Content Creation as Variation

Overview

Input: single photograph + pre-analyzed datasetInput: single photograph + pre-analyzed dataset

Page 62: Model-Driven 3D Content Creation as Variation

Overview

1. Model-driven labelled segmentation of photographed object

1. Model-driven labelled segmentation of photographed object

Page 63: Model-Driven 3D Content Creation as Variation

Overview

2. Choosing of a candidate model from the database

2. Choosing of a candidate model from the database

Page 64: Model-Driven 3D Content Creation as Variation

Overview

3. Silhouette-constrained deform-to-fit of candidate

3. Silhouette-constrained deform-to-fit of candidate

Page 65: Model-Driven 3D Content Creation as Variation

Overview

OutputOutput

Page 66: Model-Driven 3D Content Creation as Variation

Structure preservation

•Any higher-level structural info in the candidate

models is preserved during deform-to-fit

▫Symmetry relations

▫Part-level correspondence in the set

▫Controller structures [Zheng et al. @ HKUST, EG 11]

Page 67: Model-Driven 3D Content Creation as Variation

Structure preservation

•Any higher-level structural info in the candidate

models is preserved during deform-to-fit

▫Symmetry relations

▫Part-level correspondence in the set

▫Controller structures [Zheng et al. @ HKUST, EG 11]

•Structures also serve to constrain deformation of

candidate model

Page 68: Model-Driven 3D Content Creation as Variation

Controller representations

•Controllers: cuboids and generalized cylinders

•Relations: symmetry, proximity, etc.

Fitting primitivesFitting primitives

Page 69: Model-Driven 3D Content Creation as Variation

Controller representations

•Controllers: cuboids and generalized cylinders

•Relations: symmetry, proximity, etc.

Fitting primitivesFitting primitives

Page 70: Model-Driven 3D Content Creation as Variation

Deformation of controllers

photophoto

Page 71: Model-Driven 3D Content Creation as Variation

Controller primitivesController primitives

Deformation of controllers

photophoto candidate modelcandidate model

Page 72: Model-Driven 3D Content Creation as Variation

Controller primitivesController primitives

Deformation of controllers

Result of silhouette-driven deform-to-fit

Result of silhouette-driven deform-to-fit

photophoto candidate modelcandidate model

Page 73: Model-Driven 3D Content Creation as Variation

Structure preservation at work

symmetrysymmetry

Page 74: Model-Driven 3D Content Creation as Variation

Structure preservation at work

symmetrysymmetry

proximityproximity

Page 75: Model-Driven 3D Content Creation as Variation

Structure preservation at work

symmetrysymmetry

proximityproximity

optimizationoptimization

Page 76: Model-Driven 3D Content Creation as Variation

Structure preservation at work

symmetrysymmetry

proximityproximity

optimizationoptimization

outputoutput

Short videoShort video

Page 77: Model-Driven 3D Content Creation as Variation

Results

•Guidance in single view but coherent 3D results

Page 78: Model-Driven 3D Content Creation as Variation

Results

Page 79: Model-Driven 3D Content Creation as Variation

The Google chair challenge

Page 80: Model-Driven 3D Content Creation as Variation

Not just chairs …

Page 81: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Photos: immensely rich source of inspiration

▫Silhouette-driven deformation

▫Variation is less “intrusive” to retain high-level info of source model readily usable

Page 82: Model-Driven 3D Content Creation as Variation

Pros and cons

•Pros:▫Photos: immensely rich source of inspiration

▫Silhouette-driven deformation

▫Variation is less “intrusive” to retain high-level info of source model more readily usable

•Cons▫Variation does not create new structures

Page 83: Model-Driven 3D Content Creation as Variation

Future work

•Photo-inspired model deformation only a start

•Further model refinement, e.g., via sketches

Page 84: Model-Driven 3D Content Creation as Variation

Future work

•Photo-inspired model deformation only a start

•Further model refinement, e.g., via sketches

•Model-driven structure modification

Page 85: Model-Driven 3D Content Creation as Variation

Future work

•Photo-inspired model deformation only a start

•Further model refinement, e.g., via sketches

•Model-driven structure modification

•Other inspirations for 3D content creation

▫Sketch-inspired model variation

Page 86: Model-Driven 3D Content Creation as Variation

Future work

•Photo-inspired model deformation only a start

•Further model refinement, e.g., via sketches

•Model-driven structure modification

•Other inspirations for 3D content creation

▫Sketch-inspired model variation

•Style transfer with unknown style in a set

Page 87: Model-Driven 3D Content Creation as Variation

Thank you, 谢谢

TAUTAU ZJUZJUNUDTNUDTSFUSFU