hRecent Research Summary(2004-2008)(2004-2008)

W i ChWei Chen chenwei@cad.zju.edu.cnhttp://www cad zju edu cn/~chenweihttp://www.cad.zju.edu.cn/~chenweiZhejiang University

Contents (2004-2008)

Photorealistic rendering of landscapeg p

GPU-based volume visualization

Stroke-based accurate ROI extraction

Domain-driven visual analytics

Example-based illustration

Large-scale landscape simulation Wetland of West Lake, Hangzhou, g

Hardware-Accelerated Adaptive EWA Splatting

QualityQuality Westover1989Westover1989

EWAEWACrawfis 1993Crawfis 1993

Swan 1997Swan 1997

ImageImage--alignedaligned

Mueller 1999Mueller 1999

Huang 2000Huang 2000

SwanSwan

Zwicker 2001Zwicker 2001

Xue 2003Xue 2003

Our workOur work

AxisAxis--

alignedaligned

SpeedSpeedSoftwareSoftware Texture splatsTexture splats Fast Fast splatssplats OpenGL exOpenGL ex

Framework for point and volume splatting

L R lid Low-passfilter

Resamplingfilter

Warped recon-struction kernel

if depth > A use low-pass filter

if depth < Bif depth < Buse reconstruction filter

Hardware-Accelerated Adaptive EWA Splatting Conservative VoxelizationConservative correctness: all voxels

intersecting the input model are recognized.

Previous approach: generate a single voxel for each pixel by using the depth in the pixel center

Our approach: generate multiple voxels for each pixel by computing the depth range in the pixelusing the depth in the pixel center the depth range in the pixel

Conservative Voxelization Easy Image Matting

User draws strokes on imageKnown propagates to unknownTwo lineally solvable energe itemsTwo lineally solvable energe itemsLocal refinement

Iterative Optimization

Initial Input Final Matte

Results

BayesianKnockout 2PoissonInput image Trimap

Strokes BP Matting Global Easy Matting

Video Demo

Interactive Space-Time Video Matting

User draws strokes on video cubeA new volume expansion schemeExtract the matte and foreground over g

the 3D video volume by optimization

Interactive Demo Background estimationInteractive Demo Background estimation

Skate

Amira

Eli b thElizabeth

Kim

FiFire

Easy Mesh Matting

User draws strokes on meshWorks for overlaying surface or fuzzy

boundaryy

Visual Analytics for flow features on unstructured grids

Visual Analytics for flow features on unstructured grids Visual Analytics of GCxGC-TOF for cancer detection

• OMICS: Genomics: SNP; Lipidomics: 1D; Proteomics: MS (2D)Metabolomics: GCxGC MS / NMR (2D)( )

2D/3D comparative visualization Comparative visualization and exploration of potential bio-markers

Live demo

Motivation

Illustration is a visual representation

Example-based Illustration

Learn from examples for case that are diffi lt t t d d ldifficult to represent and model Existing 2D illustrations Existing models and datasets

Fulfill example based illustration by means of Fulfill example-based illustration by means of

Shape deformation Texture synthesis

Overview

Shape and shape variations by examples

C f Convey objects from measured datasets Interactive shape manipulation Example-based shape transfer

Appearance and rendering styles by examples

Texture synthesis and transfer Texture synthesis and transfer Rendering styles by examples

Computer-generated illustration

Data Acquisition Data Manipulation

Interactive Illumination3D Analogy Illumination3D Analogy

3D Deformation

2D Analogy

2D Deformation

2D Analogy

DTI-based Fiber Illustration

Proximity threshold

Hierarchical clustering

Parameter Setting g

proximity threshold =1.5 mm 2.5 mm 3.5 mm 4.5 mm 5.5 mmp y

cluster size > 10

Smoothed geometric hulls

A fiber cluster Alpha shape

After smoothing Fiber cluster + smoothed alpha shape

Principal fibers

Several principal fibers

Abstractive Exploration of Hierarchical Fibers DTI-based Muscle Illustration

Example based shape transfer

2D to 3D Using curve to drive deformation

Live demo

Example based shape transfer

2D to 3D Using distance field to convert surface to data

to get smooth boundary effects

Example based shape transfer

3D stylization from 2D example Context curves, silhouette, feature points,

local geometric details

Video demo

Example based shape transfer

2D deformation by example Differential based 2D mesh manipulation

Example based shape transfer

2D deformation by example Example-based shape manipulation

Example based shape transfer

2D deformation by examples Flexible post-process to modify the results

Video demo

Stippling by examples

Image space stippling by examples GLCM (gray-level co-occurrance matrix)

Stippling by examples Stippling by examples

Acknowledgments

Collaborators:D id S Eb t Q h P S Zh St f C i Aid David S. Ebert, Qunsheng Peng, Song Zhang, Stefan Correria, Aidong Lu, Dan Raftery, Ligang Liu, Xiao Liang, Yu Guan, Zi’ang Ding, Colin Zheng, Long Zhang, Colin Zheng, Yuyang Song, Kelly Gaither, Hanspeter Pfister Liu Ren Mattias Zwicker Zhao Dong Hujun BaoHanspeter Pfister, Liu Ren, Mattias Zwicker, Zhao Dong, Hujun Bao

Funding:Funding: NSFC: (No.60503056, 60503050, 863 program: (No. 2006AA01Z314) NewStar Project of Zhejiang University Nvidia for equipment donations. The NOAA Northern Gulf Cooperative Institute (NA06OAR4320264

06111039), the Research Initiation Program, Mississippi State University, the NIA PAR-03-056, the Alzheimer’s Association (NIRG-03-6195), the Ittleson Fund, Brown University, NSF Grants 0081581, NSF Grants 0081581, 0121288, 0328984, and the U.S. Department of Homeland Security.

chen23@purdue educhen23@purdue.eduhttp://www.cad.zju.edu.cn/home/chenwei