Advances In Computer Graphics 24th Computer Graphics International Conference Cgi 2006 Hangzhou China June 2628 2006 Proceedings 1st Edition Monssef Alsweis

Advances In Computer Graphics 24th Computer
Graphics International Conference Cgi 2006
Hangzhou China June 2628 2006 Proceedings 1st
Edition Monssef Alsweis download
https://ebookbell.com/product/advances-in-computer-graphics-24th-
computer-graphics-international-conference-cgi-2006-hangzhou-
china-june-2628-2006-proceedings-1st-edition-monssef-
alsweis-1547302
Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Advances In Computer Graphics 39th Computer Graphics International
Conference Cgi 2022 Virtual Event September 1216 2022 Proceedings
Nadia Magnenatthalmann
https://ebookbell.com/product/advances-in-computer-graphics-39th-
computer-graphics-international-conference-cgi-2022-virtual-event-
september-1216-2022-proceedings-nadia-magnenatthalmann-48839286
Advances In Computer Graphics 40th Computer Graphic International
Conference Cgi 2023 Shanghai China August 28 September 1 2023
Proceedings Part I Bin Sheng
https://ebookbell.com/product/advances-in-computer-graphics-40th-
computer-graphic-international-conference-cgi-2023-shanghai-china-
august-28-september-1-2023-proceedings-part-i-bin-sheng-54936436
Advances In Computer Graphics 37th Computer Graphics International
Conference Cgi 2020 Geneva Switzerland October 2023 2020 Proceedings
1st Ed Nadia Magnenatthalmann
https://ebookbell.com/product/advances-in-computer-graphics-37th-
computer-graphics-international-conference-cgi-2020-geneva-
switzerland-october-2023-2020-proceedings-1st-ed-nadia-
magnenatthalmann-22496440
Advances In Computer Graphics 38th Computer Graphics International
Conference Cgi 2021 Virtual Event September 610 2021 Proceedings Nadia
Magnenatthalmann
https://ebookbell.com/product/advances-in-computer-graphics-38th-
computer-graphics-international-conference-cgi-2021-virtual-event-
september-610-2021-proceedings-nadia-magnenatthalmann-35997518

Advances In Computer Graphics And Computer Vision International
Conferences Visapp And Grapp 2006 Setubal Portugal February 2528 2006
Revised Selected Papers Jose Braz Et Al
https://ebookbell.com/product/advances-in-computer-graphics-and-
computer-vision-international-conferences-visapp-and-
grapp-2006-setubal-portugal-february-2528-2006-revised-selected-
papers-jose-braz-et-al-4096692
Advances In Computer Graphics 36th Computer Graphics International
Conference Cgi 2019 Calgary Ab Canada June 1720 2019 Proceedings 1st
Ed Marina Gavrilova
https://ebookbell.com/product/advances-in-computer-graphics-36th-
computer-graphics-international-conference-cgi-2019-calgary-ab-canada-
june-1720-2019-proceedings-1st-ed-marina-gavrilova-10488854
Advanced Methods In Computer Graphics With Examples In Opengl Mukundan
https://ebookbell.com/product/advanced-methods-in-computer-graphics-
with-examples-in-opengl-mukundan-21995520
Computer Graphics In Python Advanced Vector Graphics Using Pycairo And
Python Martin Mcbride
https://ebookbell.com/product/computer-graphics-in-python-advanced-
vector-graphics-using-pycairo-and-python-martin-mcbride-52520654
Probabilistic Graphical Models Principles And Applications Advances In
Computer Vision And Pattern Recognition Luis Enrique Sucar
https://ebookbell.com/product/probabilistic-graphical-models-
principles-and-applications-advances-in-computer-vision-and-pattern-
recognition-luis-enrique-sucar-232063682

Lecture Notes in Computer Science 4035
Commenced Publication in 1973
Founding and Former Series Editors:
Gerhard Goos, Juris Hartmanis, and Jan van Leeuwen
Editorial Board
David Hutchison
Lancaster University, UK
Takeo Kanade
Carnegie Mellon University, Pittsburgh, PA, USA
Josef Kittler
University of Surrey, Guildford, UK
Jon M. Kleinberg
Cornell University, Ithaca, NY, USA
Friedemann Mattern
ETH Zurich, Switzerland
John C. Mitchell
Stanford University, CA, USA
Moni Naor
Weizmann Institute of Science, Rehovot, Israel
Oscar Nierstrasz
University of Bern, Switzerland
C. Pandu Rangan
Indian Institute of Technology, Madras, India
Bernhard Steffen
University of Dortmund, Germany
Madhu Sudan
Massachusetts Institute of Technology, MA, USA
Demetri Terzopoulos
University of California, Los Angeles, CA, USA
Doug Tygar
University of California, Berkeley, CA, USA
Moshe Y. Vardi
Rice University, Houston, TX, USA
Gerhard Weikum
Max-Planck Institute of Computer Science, Saarbruecken, Germany

Tomoyuki Nishita Qunsheng Peng
Hans-Peter Seidel (Eds.
Advancesin
ComputerGraphics
24th Computer Graphics International Conference, CGI 2006
Hangzhou, China, June 26-28, 2006
Proceedings
13

Volume Editors
Tomoyuki Nishita
University of Tokyo, Graduate School of Frontier Sciences
Department of Complexity Science and Engineering
Tokyo, Japan
E-mail: [email protected]
Qunsheng Peng
Zhejiang University, State Key Lab of CAD and CG
Hangzhou, China
E-mail: [email protected]
Hans-Peter Seidel
Max-Planck Institute for Informatics
Stuhlsatzenhausweg 85, 66123 Saarbrücken, Germany
E-mail: [email protected]
Library of Congress Control Number: 2006927817
CR Subject Classiﬁcation (1998
LNCS Sublibrary: SL 6 – Image Processing, Computer Vision, Pattern Recognition,
and Graphics
ISSN 0302-9743
ISBN-10 3-540-35638-X Springer Berlin Heidelberg New York
ISBN-13 978-3-540-35638-7 Springer Berlin Heidelberg New York
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, speciﬁcally the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting,
reproduction on microﬁlms or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965,
in its current version, and permission for use must always be obtained from Springer. Violations are liable
to prosecution under the German Copyright Law.
Springer is a part of Springer Science+Business Media
springer.com
© Springer-Verlag Berlin Heidelberg 2006
Printed in Germany
Typesetting: Camera-ready by author, data conversion by Scientiﬁc Publishing Services, Chennai, India
Printed on acid-free paper SPIN: 11784203 06/3142 543210

Preface
The 24
th
Computer Graphics International Conference (CGI 2006) was held during
June 26–28, 2006, in Hangzhou, China. This volume contains 39 full papers and 39
short papers accepted by CGI 2006. CGI conference was initially founded by the
Computer Graphics Society in 1983 and has now become a widely recognized,
high-quality academic conference in the field of computer graphics. Recent CGI
conferences were held in New York (2005
(2002), Hong Kong (2001
The CGI 2006 Program Committee received an overwhelming 387 submissions
from many countries worldwide. China and Korea contributed many enthusiastic
submissions. Based on the strict review comments of international experts, we selected
38 full papers and 37 short papers for presentations.
The main topics covered by the papers in this volume include:
• Digital geometry processing and meshes
• Physically based animation
• Figure modeling and animation
• Geometric computing and processing
• Non-photorealistic rendering
• Image-based techniques
• Visualization
We are grateful to all the authors who submitted their papers to CGI 2006, to the
international Program Committee members and external reviewers for their valuable
time and effort spent in the review process, and members of the Organizing Committee
for their hard work which made this conference successful. Finally, we would like to
thank the National Natural Science Foundation of China and K. C. Wong Education
Foundation, Hong Kong, for their financial support.
Hans-Peter Seidel, Tomoyuki Nishita, Qunsheng Peng
Program Co-chairs
Nadia Magnenat-Thalmann, Yunhe Pan, Heung-Yeung Shum
Conference Co-chairs

Committee List
Conference Co-chairs
Nadia Magnenat-Thalmann (University of Geneva, Switzerland)
Yunhe Pan (Zhejiang University, China)
Heung-Yeung Shum (Microsoft Research Asia)
Program Co-chairs
Hans Peter Seidel (MPI, Germany)
Tomoyuki Nishita (Tokyo University, Japan)
Qunsheng Peng (Zhejiang University, China)
Organizing Committee Co-chairs
Jiaoying Shi (Zhejiang University, China)
George Baciu (Hong Kong Polytechnic University, China)
Hanqiu Sun (The Chinese University of Hong Kong, China)
Organizing Committee
Xiaogang Jin, Wei Chen, Hongxin Zhang, Qi Shen, Jinhui Yu, Min Hu
(Zhejiang University, China)
Publication Chair
Hongxin Zhang (Zhejiang University, China)
Treasurer Chair
Min Hu (Zhejiang University, China)
Program Committee
Ken Anjyo (OLM Digital, Inc., Japan)
George Baciu (Hong Kong Polytechnic University, China)
Norman Badler (University of Pennsylvania, USA)
Hujun Bao (Zhejiang University, China)
Dominique Bechmann (Louis Pasteur University Strasbourg, France)
Kadi Bouatouch (University of Rennes, France)

VIII Organization
Pere Brunet (Technical University of Catalonia, Spain)
Wei Chen (Zhejiang University, China)
Fuhua (Frank) Cheng (University of Kentucky, USA)
Daniel Cohen-Or (Tel Aviv University, Israel)
Frederic Cordier (KAIST, Korea)
Guozhong Dai (The Chinese Academy of Sciences, China)
Oliver Deussen (University of Konstanz, Germany)
Yoshinori Dobashi (Hokkaido University, Japan)
Gershon Elber (Israel Institute of Technology, Israel)
Thomas Ertl (University of Stuttgart, Germany)
Shiaofen Fang (Indiana University-Purdue University Indianapolis, USA)
Dieter Fellner (Graz University of Technology, Austria)
Robin Forrest (University of East Anglia, UK)
Issei Fujishiro (Tohoku University, Japan)
Martin Goebel (GMD, Germany
Michael Goesele (University of Washington, USA)
Mark Green (UOIT, Canada)
Xianfeng Gu (Stony Brook University, USA)
Baining Guo (Microsoft Research Asia)
Wolfgang Heidrich (University of British Columbia, Canada)
Shimin Hu (Tsinghua University, China)
Horace H.S. Ip (City University of Hong Kong, China)
Ioannis Ivrissimtzis (Coventry University, UK)
Xiaogang Jin (Zhejiang University, China)
Tao Ju (Washington University in St. Louis, USA)
Kazufumi Kaneda (Hiroshima University, Japan)
Sing Bing Kang (Microsoft Research Asia)
Hyung Woo Kang (University of Missouri, USA)
Jan Kautz (MIT, USA
Deok-Soo Kim (Hanyang University, Korea)
Kujin Kim (Kyungpook National University, Korea)
Myoung Jun Kim (Ewha Womans University, Korea)
Myung-Soo Kim (Seoul National University, Korea)
Hyeong-Seok Ko (Seoul National University, Korea)
Leif P. Kobbelt (RWTH Aachen, Germany)
Rynson Lau (City University of Hong Kong, China)
Jehee Lee (Seoul National University, Korea)
Seungyong Lee (Pohang University of Science and Technology, Korea)
Hendrik Lensch (Stanford University, USA)
Zicheng Liu (Microsoft Research, USA)
Kwan-Liu Ma (University of California, USA)
Weiyin Ma (City University of Hong Kong, China)
Nadia Magnenat-Thalmann (University of Geneva, Switzerland)
Marcus Magnor (Technical University Braunschweig, Germany)
Xiaoyang Mao (Yamanashi University, Japan)
Ahmad Nasri (American University of Beirut, Lebanon)

Organization IX
Miguel Otaduy (ETH-Zurich, Switzerland)
Ming Ouhyoung (National Taiwan University, China)
Zhigeng Pan (Zhejiang University, China)
Jean-Claude Paul (CNRS, France)
Les Piegl (University of South Florida, USA)
Werner Purgathofer (Vienna University of Technology, Austria)
Dongxu Qi (Macao University of Science and Technology, China)
Hong Qin (State University of New York, USA)
Stephane Redon (INRIA, France)
Jon G. Rokne (University of Calgary, Canada)
Lawrence Rosenblum (Naval Research Laboratory, USA)
Bodo Rosenhahn (Max Planck Center, Germany)
Dimitris Samaras (State University of New York, USA)
Hyewon Seo (Computer Graphics and Applications Lab at CNU, Korea)
Jiaoying Shi (Zhejiang University, China)
Hyunjoon Shin (Ajou University, Korea)
Sung Yong Shin (KAIST, Korea)
Yeong-Gil Shin (Seoul National University, Korea)
Yoshihisa Shinagawa (University of Illinois, USA)
Harry Shum (Microsoft Research Asia)
Claudio Silva (University of Utah, USA)
Michela Spagnuolo (IMATI Genova, Italy)
Jos Stam (Alias Systems, USA)
Hanqiu Sun (The Chinese University of Hong Kong, China)
Chiew-Lan Tai (Hong Kong University of Science and Technology, China)
Kai Tang (Hong Kong University of Science and Technology, China)
Demetri Terzopoulos (University of Toronto, Canada)
Daniel Thalmann (Swiss Federal Institute of Technology, Switzerland)
Holger Theisel (MPI, Germany
Xing Tong (Microsoft Research Asia)
George Turkiyyah (University of Washington, USA)
Amitabh Varshney (University of Maryland, USA)
Wenping Wang (University of Hong Kong, Hong Kong)
Joe Warren (Rice University, USA)
Tien Tsin Wong (The Chinese University of Hong Kong, China)
Enhua Wu (The Chinese Academy of Sciences, China)
Brian Wyvill (University of Calgary, Canada)
Geoff Wyvill (University of Otago, New Zealand)
Yingqing Xu (Microsoft Research Asia)
Ruigang Yang (University of Kentucky, USA)
Steve S.N. Yang (National Tsing Hua University, China)
Jingyi Yu (University of Delaware, USA)
Yizhou Yu (University of Illinois, USA)
Hongxin Zhang (Zhejiang University, China)

X Organization
Additional Reviewer List
Yiyu Cai (Nanyang Technological University , Singapore)
Weiqun Cao (Beijing Forestry University, China)
Baoquan Chen (University of Minnesota at Twin Cities, USA)
Yanyun Chen (Microsoft Research Asia)
Falai Chen (University of Science and Technology of China, China)
Bingyu Chen (National Taiwan University, Taiwan)
Sung Woo Choi (Korea University, Korea)
Yung-Yu Chuang (University of Washington, USA)
Gordon Clapworthy (University of Luton, UK)
Shihai Dong (Peking University, China)
Jieqing Feng (Zhejiang University, China)
Shuming Gao (Zhejiang University, China)
Weidong Geng (Zhejiang University, China)
Craig Gotsman( Harvard University, USA)
Eran Guendelman (Stanford University, USA)
Yongjun Hai (Chongqing University of Posts and Telecommunications, China)
Xuli Han (Central South University, China)
Mark Harris (NVIDIA Corporation, USA)
Yuanjun He (Shanghai Jiaotong University, China)
Pheng-Ann Heng (The Chinese University of Hong Kong, China)
Wei Hua (Zhejiang University, China)
Zhiyong Huang (National University of Singapore, Singapore)
Zhongding Jiang (Fudan University, China)
Arie Kaufman (State University of New York at Stony Brook, USA)
Yingling Ke (Zhejiang University, China)
John Keyser (Texas A&M University, USA)
Yoshifumi Kitamura (Osaka University, Japan)
Shigeru Kuriyama (Toyohashi University of Technology, Japan)
Guiqing Li (South China University of Technology, China)
Sikun Li (National University of Defense Technology, China)
Hua Li (The Chinese Academy of Sciences, China)
Youdong Liang (Zhejiang University, China)
Xueyan Lin (Tsinghua University, China)
Hongwei Lin (Zhejiang University, China)
Xuehui Liu (The Chinese Academy of Sciences, China)
Yongkui Liu (Dalian Nationalities University, China)
Dingyuan Liu (Fudan University, China)
Xinguo Liu (Microsoft Research Asia)
Yanxi Liu (Carnegie Mellon University, USA)
Ligang Liu (Zhejiang University, China)
Yusheng Liu (Zhejiang University, China)
Hanqing Lu (The Chinese Academy of Sciences, China)
Lizhuang Ma (Shanghai Jiao Tong University, China)
Xiaohu Ma (Soochow University, China)

Organization XI
Huadong Ma (Beijing University of Posts and Telecommunications, China)
Ralph Martin (Cardiff University, UK)
Simon Masnou (Université Pierre-et-Marie-Curie, France)
Radomír Mech (University of Calgary, Canada)
Yutaka Ohtake (Max-Planck-Institut fur Informatik, Germany)
Rick Parent (Ohio State University, USA)
Kaihuai Qin (Tsinghua University, China)
Xueyin Qin (Zhejiang University, China)
Erik Reinhard (University of Bristol, UK)
Zen Chung Shih (National Chiao Tung University, China)
Huahao Shou (Zhejiang University of Technology, China)
Vaclav Skala (University of West Bohemia, Czech Republic)
Jizhou Sun (Tianjin University, China)
Zheng Tan (Xi’an Jiaotong University, China)
Zesheng Tang (Macao University of Science and Technology, China)
Chi-Keung Tang (The Hong Kong University of Science and Technology, China)
Frank Van Reeth (Limburgs Universitair Centrum, Belgium)
Justin W.L. Wan (University of Waterloo, Canada)
Huagen Wan (Zhejiang University, China)
Wencheng Wang (The Chinese Academy of Sciences, China)
Huamin Wang (Georgia Institute of Technology, USA)
Lifeng Wang (Microsoft Research Asia)
Jiaye Wang (Shandong University, China)
Zhaoqi Wang (The Chinese Academy of Sciences, China)
Niniane Wang (Google Inc., USA)
Guojin Wang (Zhejiang University, China)
Jin Wang (Zhejiang University, China)
Kelly Ward (The University of North Carolina at Chapel Hill, USA)
Liyi Wei (Microsoft Research Asia)
Xiaogang Xu (Dalian Naval Academy, China)
Guangyou Xu (Tsinghua University, China)
Guoliang Xu (The Chinese Academy of Sciences, China)
Dan Xu (Yunnan University, China)
Xunnian Yang (Zhejiang University, China)
Xiuzi Ye (Zhejiang University, China)
Junhai Yong (Tsinghua University, China)
Jinhui Yu (Zhejiang University, China)
Richard Zhang (Simon Fraser University, Canada)
Shusheng Zhang (Northwestern Polytechnic University, China)
Caiming Zhang (Shandong Economic University, China)
Jian J. Zhang (Bournemouth University, UK)
Jiwen Zhang (Zhejiang University, China)
Hongkai Zhao (University of California, USA)
Jianmin Zheng (Nanyang Technological University, Singapore)
Kun Zhou (Microsoft Research Asia)
Miaoliang Zhu (Zhejiang University, China)

Table of Contents
Regular Papers
Wang-Tiles for the Simulation and Visualization of Plant Competition
Monssef Alsweis, Oliver Deussen.................................1
Multi-layered Stack Mosaic with Rotatable Objects
Jin Wan Park, Kyung Hyun Yoon, Seung Taek Ryoo...............12
Appearance and Geometry Completion with Constrained Texture
Synthesis
Chunxia Xiao, Wenting Zheng, Yongwei Miao, Yong Zhao,
Qunsheng Peng................................................24
Highly Stylised Drawn Animation
Fabian Di Fiore, Frank Van Reeth, John Patterson, Philip Willis....36
Non-uniform Diﬀerential Mesh Deformation
Dong Xu, Hongxin Zhang, Hujun Bao............................54
Skeleton-Based Shape Deformation Using Simplex Transformations
Han-Bing Yan, Shi-Min Hu, Ralph Martin........................66
Skeleton-Driven Animation Transfer Based on Consistent Volume
Parameterization
Yen-Tuo Chang, Bing-Yu Chen, Wan-Chi Luo, Jian-Bin Huang.....78
Sketch Based Mesh Fusion
Juncong Lin, Xiaogang Jin, Charlie C.L. Wang....................90
Real-Time Rendering of Point Based Water Surfaces
Kei Iwasaki, Yoshinori Dobashi, Fujiichi Yoshimoto,
Tomoyuki Nishita..............................................102
Controllable Multi-phase Smoke with Lagrangian Particles
Byung-Seok Roh, Chang-Hun Kim...............................115
An Approximate Image-Space Approach for Real-Time Rendering
of Deformable Translucent Objects
Yi Gong, Wei Chen, Long Zhang, Yun Zeng, Qunsheng Peng........124

XIV Table of Contents
Interactively Rendering Dynamic Caustics on GPU
Baoquan Liu, Enhua Wu, Xuehui Liu............................136
Fuzziness Driven Adaptive Sampling for Monte Carlo Global
Illuminated Rendering
Qing Xu, Mateu Sbert, Zhigeng Pan, Wei Wang, Lianping Xing.....148
Manifold Parameterization
Lei Zhang, Ligang Liu, Zhongping Ji, Guojin Wang................160
Sub-sampling for Eﬃcient Spectral Mesh Processing
Rong Liu, Varun Jain, Hao Zhang...............................172
Active Contours with Level-Set for Extracting Feature Curves from
Triangular Meshes
Kyungha Min, Dimitris N. Metaxas, Moon-Ryul Jung..............185
A Feature-Preserving and Volume-Constrained Flow for Fairing
Irregular Meshes
Chunxia Xiao, Shu Liu, Qunsheng Peng, A.R. Forrest..............197
Matching 2D Shapes Using U Descriptors
Zhanchuan Cai, Wei Sun, Dongxu Qi............................209
Electric Field Force Features-Harmonic Representation for 3D Shape
Similarity
Yujie Liu, Zongmin Li, Hua Li..................................221
A Novel Data Hiding Algorithm Using Normal Vectors of 3D Model
Chung-Hsien Chang, Chung-Ming Wang, Yuan-Yu Tsai,
Yu-Ming Cheng................................................231
Shape Matching Based on Fully Automatic Face Detection on
Triangular Meshes
Wolfram von Funck, Holger Theisel, Hans-Peter Seidel.............242
Skin Color Analysis in HSV Color Space and Rendering with Fine Scale
Skin Structure
Dae Hyun Kim, Myoung-Jun Kim...............................254
Comprehending and Transferring Facial Expressions Based on
Statistical Shape and Texture Models
Pengcheng Xi, Won-Sook Lee, Gustavo Frederico, Chris Joslin,
Lihong Zhou..................................................265

Table of Contents XV
Real-Time Facial Expression Mapping for High Resolution 3D Meshes
Mingli Song, Zicheng Liu, Baining Guo...........................277
A Comparison of Three Techniques to Interact in Large Virtual
Environments Using Haptic Devices with Limited Workspace
Lionel Dominjon, Anatole L´ecuyer, Jean-Marie Burkhardt,
Simon Richir..................................................288
Trajectory-Based Grasp Interaction for Virtual Environments
Zhenhua Zhu, Shuming Gao, Huagen Wan, Wenzhen Yang..........300
Research on User-Centered Design and Recognition Pen Gestures
Feng Tian, Tiegang Cheng, Hongan Wang, Guozhong Dai..........312
Simulating Pedestrian Behavior with Potential Fields
F´abio Dapper, Edson Prestes, Marco A.P. Idiart,
Luciana P. Nedel..............................................324
Providing Full Awareness to Distributed Virtual Environments Based
on Peer-to-Peer Architectures
P. Morillo, W. Moncho, J.M. Ordu˜na,J.Duato...................336
Motion Editing with the State Feedback Dynamic Model
Dengming Zhu, Zhaoqi Wang, Shihong Xia........................348
Content-Based Human Motion Retrieval with Automatic Transition
Yan Gao, Lizhuang Ma, Yiqiang Chen, Junfa Liu..................360
MIP-Guided Vascular Image Visualization with Multi-Dimensional
Transfer Function
Ming-Yuen Chan, Yingcai Wu, Huamin Qu, Albert C.S. Chung,
Wilbur C.K. Wong.............................................372
Automatic Foreground Extraction of Head Shoulder Images
Jin Wang, Yiting Ying, Yanwen Guo, Qunsheng Peng..............385
Direct Volume Rendering of Volumetric Protein Data
Min Hu, Wei Chen, Tao Zhang, Qunsheng Peng...................397
Subdivision Depth Computation for Extra-Ordinary Catmull-Clark
Subdivision Surface Patches
Fuhua (Frank) Cheng, Gang Chen, Jun-Hai Yong..................404
An Approach for Embedding Regular Analytic Shapes with
Subdivision Surfaces
Abdulwahed Abbas, Ahmad Nasri, Weiyin Ma.....................417

XVI Table of Contents
Adaptive Point-Cloud Surface Interpretation
Q. Meng, B. Li, H. Holstein....................................430
An Accurate Vertex Normal Computation Scheme
Huanxi Zhao, Ping Xiao........................................442
Short Papers
A Visibility-Based Automatic Path Generation Method for Virtual
Colonoscopy
Jeongjin Lee, Moon Koo Kang, Yeong Gil Shin....................452
Dynamic Medial Axes of Planar Shapes
Kai Tang, Yongjin Liu.........................................460
Steganography on 3D Models Using a Spatial Subdivision Technique
Yuan-Yu Tsai, Chung-Ming Wang, Yu-Ming Cheng,
Chung-Hsien Chang, Peng-Cheng Wang..........................469
Addressing Scalability Issues in Large-Scale Collaborative Virtual
Environment
Qingping Lin, Liang Zhang, Norman Neo, Irma Kusuma...........477
Symmetric Tiling Patterns with the Extended Picard Group in
Three-Dimensional Space
Rui-Song Ye, Jian Ma, Hui-Liang Li.............................486
An Eﬃcient Keyframe Extraction from Motion Capture Data
Jun Xiao, Yueting Zhuang, Tao Yang, Fei Wu.....................494
Visualization of Whole Genome Alignment with LOD Representation
Hee-JeongJin,Hwan-GueCho ..................................502
Steganography for Three-Dimensional Models
Yu-Ming Cheng, Chung-Ming Wang, Yuan-Yu Tsai,
Chung-Hsien Chang, Peng-Cheng Wang..........................510
Feature Sensitive Out-of-Core Chartiﬁcation of Large Polygonal
Meshes
Sungyul Choe, Minsu Ahn, Seungyong Lee........................518
Simulating Reactive Motions for Motion Capture Animation
Bing Tang, Zhigeng Pan, Le Zheng, Mingmin Zhang...............530

XVIII Table of Contents
Real-Time Simulation of Dynamic Mirage Scenes
Changbo Wang, Zhangye Wang, Qi Zhou,
Zhidong Jin, Qunsheng Peng....................................647
Improving the Interval Ray Tracing of Implicit Surfaces
Jorge Fl´orez, Mateu Sbert, Miguel A. Sainz,
Josep Veh´ı....................................................655
Algorithms for Vector Graphic Optimization and Compression
Mingkui Song, Richard R. Eckert, David A. Goldman..............665
Detail-Preserving Local Editing for Point-Sampled Geometry
Yongwei Miao, Jieqing Feng, Chunxia Xiao, Hui Li,
Qunsheng Peng................................................673
Automatic Stained Glass Rendering
Vidya Setlur, Stephen Wilkinson.................................682
Vision-Based Augmented Reality Visual Guidance with Keyframes
Timothy S.Y. Gan, Tom W. Drummond..........................692
Optimized Framework for Real Time Hair Simulation
Rajeev Gupta, Melanie Montagnol, Pascal Volino,
Nadia Magnenat-Thalmann.....................................702
Optimizing Mesh Construction for Quad/Triangle Schemes
Koen Beets, Johan Claes, Frank Van Reeth........................711
Rendering Optical Eﬀects Based on Spectra Representation
in Complex Scenes
Weiming Dong................................................719
GVF-Based Transfer Functions for Volume Rendering
Shaorong Wang, Hua Li........................................727
Quasi-physical Simulation of Large-Scale Dynamic Forest Scenes
Long Zhang, Chengfang Song, Qifeng Tan, Wei Chen,
Qunsheng Peng................................................735
Properties of G1 Continuity Conditions Between Two B-Spline
Surfaces
Nailiang Zhao, Weiyin Ma......................................743

Table of Contents XVII
Real-Time Shadow Volume Algorithm for Subdivision Surface Based
Models
Min Tang, Jin-Xiang Dong, Shang-Ching Chou....................538
Human Animation from 2D Correspondence Based on Motion Trend
Prediction
Li Zhang, Ling Li..............................................546
A Straightforward and Intuitive Approach on Generation and Display
of Crack-Like Patterns on 3D Objects
Hsien-Hsi Hsieh, Wen-Kai Tai..................................554
Near-Optimum Adaptive Tessellation of General Catmull-Clark
Subdivision Surfaces
Shuhua Lai, Fuhua (Frank) Cheng...............................562
Spline Thin-Shell Simulation of Manifold Surfaces
Kexiang Wang, Ying He, Xiaohu Guo, Hong Qin..................570
Target Shape Controlled Cloud Animation
Shengjun Liu, Xiaogang Jin, Charlie C.L. Wang...................578
Plausible Locomotion for Bipedal Creatures Using Motion Warping
and Inverse Kinematics
Guillaume Nicolas, Franck Multon, Gilles Berillon,
Francois Marchal..............................................586
Aerial Image Relighting: Simulating Time of Day Variations
Kartik Chandra, Neeharika Adabala, Kentaro Toyama..............594
Compression of Complex Animated Meshes
Rachida Amjoun, Ralf Sondershaus, Wolfgang Straßer..............606
A Video-Driven Approach to Continuous Human Motion Synthesis
Rongrong Wang, Xianjie Qiu, Zhaoqi Wang, Shihong Xia...........614
Spatio-temporal Visualization of Battleﬁeld Entities and Events
Qiyue Fong, Foo Meng Ng, Zhiyong Huang........................622
3D City Model Generation from Ground Images
Kyung Ho Jang, Soon Ki Jung..................................630
Anticipation Eﬀect Generation for Character Animation
Jong-Hyuk Kim, Jung-Ju Choi, Hyun Joon Shin,
In-Kwon Lee..................................................639

Table of Contents XIX
Automated Face Identiﬁcation Using Volume-Based Facial Models
Jeﬀrey Huang, Neha Maheshwari, Shiaofen Fang...................753
Feature Curves with Cross Curvature Control on Catmull-Clark
Subdivision Surfaces
Ahmad Nasri, Malcolm Sabin, Rana Abu Zaki, Nasser Nassiri,
Rami Santina.................................................761
Author Index...................................................769

Wang-Tiles for the Simulation
and Visualization of Plant Competition
Monssef Alsweis and Oliver Deussen
Department of Computer and Information Science
University of Konstanz, Germany
{Alsweis, deussen}@inf.uni-konstanz.de
http://graphics.uni-konstanz.de
Abstract.The Wang Tiles method is a successful and eﬀective tech-
nique for the representation of 2D-texture or 3D-geometry. In this paper
we present a new method to ﬁll Wang tiles with a 2D-FON distribution
or a 3D-geometry in order to achieve a more eﬃcient runtime. We extend
the Wang Tiles method to include information about their position. We
further demonstrate how the individual tiles are ﬁlled with diﬀerent in-
tensities by using the FON distribution. Additionally, we present several
new methods to eliminate errors between the tile edges and the diﬀerent
resource areas applying FON and corners relaxation techniques.
1 Introduction
Modelling and visualization of large complex scenes of plants is a diﬃcult and
time consuming task. For example, the storing of plant compositions is extremely
diﬃcult because a square kilometer of forest consists of millions of plants, hun-
dred of thousands of small trees, and numerous small scrubs. To model intricate
natural scenes, several methods already exist in computer graphics. The primary
method was introduced in [1]. A new stochastic algorithm was non-periodically
presented to cover the area with a small set of Wang Tiles [2, 3]. This method
provides Wang Tiles with the eﬃciency of being re-usable tiles, which in turn
allows for rendering larger areas of complicated textures, patterns or prelighted
geometry at more eﬃcient runtime. Wang Tiles are a set of squares in which
each edge of each tile is colored. Matching colored edges are aligned to tile an
area. In [1] a set of eight tiles were used to cover an area, and a 2D Poisson Disk
distribution was applied to ﬁll the tiles. The aim is to render beautiful and real-
istic natural scenes. We use four sets or more of tiles instead of one set. Each set
of tiles consists of eight tiles. All sets of tiles share the same color coding. How-
ever, they diﬀerentiate from each other through the intensity of the distributions
on the tiles. In the ﬁnal tiling and scene, rsp., the intensity of the distribution
depends on the terrain or on the amount of the resources in which the tiles are
positioned. For example, the intensity of the distribution of the plants in an area
with poor resources diﬀers from those in the area with rich resources. Applying
this method enables to generate large diﬀerences in the intensity of the plants in
H.-P. Seidel, T. Nishita, and Q. Peng (Eds.): CGI 2006, LNCS 4035, pp. 1–11, 2006.
cﬀSpringer-Verlag Berlin Heidelberg 2006

2 M. Alsweis and O. Deussen
two areas. To eliminate this problem, we use several sets and apply some kind
of super sampling (antialiasing) method for the selection of the matching sets.
Often plant positions form a 2D Poisson Disc Distribution: each object is
presented by a circular area in which are no other objects and in which no other
areas overlap. This is done due to what happens in nature when plants prevent
other plants to grow in their vicinity. To model this behaviour, we applied the so-
called FON (Field-Of-Neighbourhood) distribution model [4]. In the FON model
each object is presented in a circular area. The size of this circle depends of the
size of the object and of the terrain or the amount of resources.
2 Previous Works
There have been diﬀerent approaches for accelerating of rendering for complex
models. Many approaches aimed to avoid visual artifacts which have been re-
peatedly created using simple methods.
One of these approaches is the rendering of complicated natural scenes through
the application of theWang Tiling Method[1, 2, 3] in which a limited set of
tiles is used to layout a possibly inﬁnite plane. We combined this model with an
ecosystem simulation model, the so-calledFON (Field Of Neighbourhood)
Modellby Berger und Hildenbrandt in [4] and [5]. This model is an individuum-
based model and describes a circular zone of the area of inﬂuence around the
plant. The radius of this area determines the distance in which the neighboring
plants inﬂuence each other. In [6] the radius of this zone is speciﬁed by a non-
linear function which depends from the basal radius of the plant, the size of the
plant, the amount of resources needed by the plant, and the area necessary to
provide the resources.
Modelling of Complex Ecosystems:In some earlier works [7, 8, 9] nature was
rendered using artiﬁcial mechanisms. Other scientists [10, 11] tried to imitate the
actual natural processes. Both approaches are still used and many natural and
other complex phenomena were modelled in [12, 13, 14]. In [15, 16, 17] most im-
portant factors that inﬂuence the shape, the structure, and the development of a
plant were modeled. The ﬁrst work thatsimulated the competition and the de-
velopment of the plants was published by [18]. Lane and Prusinkiewicz extended
this work and developed a new method called Multiset L-Systems that was used
for the description of the competition between plants. In [19] is described how
plants grow in groups, and how the communities compete for resources. In [20], a
new method was introduced for the visualization and simulation of the develop-
ments of a plant group. In this method, the plant competition for resources was
presented as a symmetric and asymmetric competition. Asymmetric competition
takes place when plants diﬀer in size or in type.
3OurMethod
To tile an area, we apply the Wang Tiling Method. A Wang Tile set consists
of square tiles with color-coded edges. The squares cannot be rotated. A valid

Wang-Tiles for the Simulation and Visualization of Plant Competition 3
(a) (b) (c)
Fig. 1.a) Tile set; b) Valid tiling; c) four sets of Wang tiles with diﬀerent distribution
intensity
tiling of the inﬁnite plane consists of any number of copies from the set laid
down such that all continuous edges have matching colors. (s. Fig. 1).
3.1 Tiling
In this work we use four sets or more, each consisting of a set of eight tiles. Each
set resembles the other in the color at the edge of the tiles, and diﬀerentiates
itself from the others through the intensity of the distribution in the tiles. The
reason for the use of of four or more sets of tiles with diﬀerent intensity on
the edges is as follows: From Fig. 2(a) it can be seen that the area as well as
the diﬀerent levels of resources diﬀer for diﬀerent terrains. Consequently, the in
the white circles marked tiles from Fig. 2(b
although all marked tiles have the same edge color. In order to receive a soft
border between two diﬀerent resource areas, and at the same time to position
(a) (b) (c
Fig. 2.(a
resulting tiling match, diﬀerent intensities are applied; (c) a window is centered around
the sampled value (i, j); (d
value

4 M. Alsweis and O. Deussen
the tiles matching the set of tileI
ﬃ
in the area, we apply an image operator for
averaging. The intesnsityI
ﬃ
(i, j) is computed as:
1
M
Si+k
ﬀ
p=Si−k
Sj+k
ﬀ
q=Sj−k
I(p, q)h(Si−p, Sj−q)(1)
Whereby (i, j) indicates the tile in the position (i, j) (see Fig. 2(a)),Sis a
scaling factor, andhis a ﬁlter of the dimensionk. The selection of the set order
is easy to understand and to realize, however, it is computationally expensive.
In Fig. 2(a)-(b) a window is centered over an sampled value and a weighted
sum of products is achieved by means of multiplication of each sampled value
with the respective weight in the ﬁlter. The weight can be adjusted in order
to implement diﬀering ﬁlter kernels. The digital ﬁltering continues in that the
window is moved throughn=k∗ksampling value and the next weighted sum
of products is computed.
The use of a 3×3 window indicates that nine sampled values are part of the
ﬁnal tile computation. On the other hand, the use of a 7×7 windows includes
a computation of 49 integer multiplication. The result of this computational
expense is obvious. For a virtual 3×3 window ﬁlter, a ﬁlter on tiles in the
virtual image can be positioned, in that step-by-step three super-sampling tiles
are used.Mis computed by the following equation:
M=
2k
ﬀ
p
ﬀ
=0
2k
ﬀ
q
ﬀ
=0
h(p
ﬃ
,q
ﬃ
)(2)
Here,Iis the sampling value of the level of the resources in a tile. In order to
determine this value in a tile, the resources from diﬀerent positions in certain
tiles are sampled. The average of these sampled resources levels is designated to
be the sampled value.
3.2 FON Distribution
In order to ﬁll each individual tile with diﬀerent intensity, we use the already
mentioned FON Modell Distribution. In this model each individual plant has a
circular zone of inﬂuence (ZOI), the radius of the zone determines the distance
in which the individual plant inﬂuences the neighboring plant.
Each individual plant is identiﬁed in the tile through its position, size, and the
age. Additionally, each plant has a FON inﬂuence zone. To determine the FON
inﬂuence zone, a non-linear function of the basal radius is applied [4] (s. Fig. 3).
RFON=a(Rbasal)
b
(3)
RFONis the radius of the FON inﬂuence zone of the individual plant.Rbasalis
the radius of the individual plant (Fig. 3).aandbare constants and depend on
the intensity of the resources in the ground and of the intensity of the light in
above-ground. Typicallyahas the range of [12,133] andb∈[1.2,2.3].

Wang-Tiles for the Simulation and Visualization of Plant Competition 5
Fig. 3.The FON model. The zone of inﬂuence (R FON) depends on the diameter of
the trunk.
(a) (b
Fig. 4.In (a) represents the intensity of the resources. In (b) the FON-Tile Method is
applied. In (c) shows the noice between the tiles.
Application of the tiles and the FON Distribution in Fig. 4 produces the
result in subﬁgure (b). Subﬁgure (a) shows an image map of the intensity of
the resources in the ground. In (b) is shown that the intensity of the plants
in the resource-poor areas is less than in the resource-rich areas. As a result of
implementing the averaging, we can derive a gradual change of that the intensity
of the plants between two diﬀerent regions.
3.3 FON Relaxation
In Fig. 4(c
In order to eliminate this error, we apply the FON relaxation method at the
edge of the tiles.
The FON relaxation method is applied to test whether the FON-zone of plants
positioned at the edge of one tile overlap with the FON-zone of the plants posi-
tioned at the edge of the neighboring tile, and whether the FON-zone of one plant
is smaller or larger than that of the neighboring plant at a certain threshold. If
the test is positive, the plant must be removed from the respective tile.
In Fig. 5 it is shown that plant number one passed the test and has to be
removed from the tile. On the other side the test was not passed for plant
number two, and therefore it can remain in the tile. If we compare Fig. 4(c) with

6 M. Alsweis and O. Deussen
Fig. 5.The circles are presented in two conditions at the edge of the tiles. The ﬁrst
circles is smaller than the other one. The second circle is larger than the other one.
Fig. 6.As a result of implementing the FON relaxation in (c) (Fig. 4), the noise
between the tiles was removed.
Fig. 6, we note, that the distribution of the plants is now more even. However,
the price is that some small amount of computation has to be done during the
tile-based layout.
3.4 Relaxation of Corners
From Fig. 7 we note that the diﬀerence of the plant intensity between the tileA
and the tilesB,CundDcreates corners. These corners produces a non-realistic
appearance about the appearance of the plants in the scene. To eliminate this
unwanted appearance, we apply the relaxation method to tileA. The implemen-
tation of the this relaxation method is divided into the following steps:
1. It is tested whether tileAproduces corners This happens when the intensity
inAdiﬀers from the intensity inB,CundD.

Wang-Tiles for the Simulation and Visualization of Plant Competition 7
Fig. 7.The intensity corners between the tilesAand tilesB,CandD
Fig. 8.Distribution of the tilesAinto two partsUandO
2. The tileAis divided into two diagonal parts (lower partUand upper partO)
in order to be ﬁlled with two diﬀerent intensities. For example as in (Fig. 8).
3. The intensity in partUwill not be changed. against it the intensity in part
Odepends on the intensities in tilesB,CundD. The arrangements ofB,
CundDcontrol the deﬁnition in the set order in partU.
4. Fon-relaxation is applied between diagonal parts.
The application of the last four steps in Fig. 4(c
plant distribution without corners in Fig. 6.

8 M. Alsweis and O. Deussen
4 Implementation
Implementation of this method results in a good quality of a natural scene, so
that the noise between the tiles and the noise between the two diﬀerent resource
areas are improved. Additionally, this scene is rendered in real-time. To imple-
ment this method, we use a discrete system of several sets. The implementation
of these systems can be divided into the following steps:
1. We produce diﬀerent sets of tiles. Each set diﬀers in plant intensity from the
other.
2. each plant is presented in two circles. The ﬁrst represents the size of the
plants, and the second represents the FON-zone of the plant.
3. We apply the Wang-Tiling method for diﬀerent sets in order to cover the
area.
4. In order to eliminate the noise between the tiles, we apply the FON-relaxation
method.
5. In order to remove the corners between the tiles, we apply the corner relax-
ation method.
The plants used in this approach were modelled with the Xfrog Software [21],
and were then imported into our systems as POVRAY MESH2 in Fig. 9 and
Fig. 10.
Fig. 9.This image shows the rendering of a scene of a meadow and ﬂowers applying
the tiling method with diﬀerent intensities of resources
Fig. 10.In this image diﬀerent resource intensities in a ﬁeld of sunﬂowers are shown
5Results
We introduced a ﬁne eﬃcient method to render natural scenes with a more real-
istic appearance. The use of more sets of tiles with diﬀerent intensity produces

Wang-Tiles for the Simulation and Visualization of Plant Competition 9
Fig. 11.In this image diﬀerent resource intensities in African landscape are shown
Fig. 12.In this image diﬀerent resource intensities in African landscape from other
camera position are shown

10 M. Alsweis and O. Deussen
Fig. 13.In this image diﬀerent resource intensities in Amazon landscape are shown
scenes that have a more natural appearance (see Figs. 9 and 10). The imple-
mentation of the FON-relaxation and corner relaxation in one scene produces
smooth edges between the tiles and a gradual transformation between two dif-
ferent resource intensities (Figs. 11, 12 and 13).
References
1. Michael F. Cohen, Jonathan Shade, Stefan Hiller, and Oliver Deussen. Wang tiles
for image and texture generation.Proceedings of ACM SIGGRAPH., pages 287–
294, July 2003.
2. H. WANG. Proving theorems by pattern recognition ii.Bell Systems Technical
Journal., 40:142, 1961.
3. H. WANG. Games, logic, and computers.Scientiﬁc American (November)., page
98106, 1965.
4. Berger U. and Hildenbrandt H. A new appoarch to spatially explicit modelling
of forest dynamics: Spacing, ageing and neighbourhood competition of mangrove
trees.Ecological Modelling, 132:287302, 2000.
5. S. Bauer, U. Berger, H. Hildenbrandt, and V. Grimm. Cyclic dynamics in simulated
plant populations.Proc R Soc Lond B Biol Sci., 269(1508):2443–2450, 2002.
6. U. Berger, H. Hildenbrandt, and V. Grimm. Towards a standard for the individual-
based modeling of plant opulations: self-thinning and the ﬁeld of neighbourhood
approach.Nat. Ressource Model, 15:3954, 2002.

Wang-Tiles for the Simulation and Visualization of Plant Competition 11
7. Geoﬀrey Y.Gardner. Visual simulation of clouds. in proceedings of the 12th
annual conference on computer graphics and interactive techniques.ACM Press.,
page 297304, 1985.
8. Robert Marshall., Rodger Wilson., and Wayne Carlson. Procedure models for
generating three-dimensional terrain, in proceedings of the 7th annual conference
on computer graphics and interactive techniques.ACM Press., page 154162, 1980.
9. Nelson L. Max. Vectorized procedural models for natural terrain: Waves and
islands in the sunset, in proceedings of the 8th annual conference on computer
graphics and interactive techniques.ACM Press, page 317324, 1981.
10. Alain Fournier and William T. Reeves. A simple model of ocean waves. in
proceedings of the 13th annual conference on computer graphics and interactive
techniques.ACM Press, page 7584, 1986.
11. Craig W., Reynolds., Flocks, herds, and schools. A distributed behavioral model,
in proceedings of the 14th annual conference on computer graphics and interactive
techniques,.ACM Press, page 2534, 1987.
12. Yoav I., H. Parish, and Pascal Muller. Procedural modeling of cities, in proceedings
of the 28th annual conference on computer graphics and interactive techniques,.
ACM Press, page 301308, 2001.
13. Yoshinori Dobashi., Kazufumi Kaneda., Hideo Yamashita., Tsuyoshi Okita., and
Tomoyuki Nishita. A simple, eﬃcient method for realistic animation of clouds,in
proceedings of the 27th annual conference on computer graphics and interactive
techniques.ACM Press/Addison-Wesley Publishing Co, page 1928, 2000.
14. Paul Fearing. Computer modelling of fallen snow. in proceedings of the 27th
annual conference on computer graphics and interactive techniques.ACM
Press/Addison-Wesley Publishing Co, page 3746, 2000.
15. J. Arvo and D. Kirk. Modeling plants with environment-sensitive automata.In
Proceedings of Ausgraph88, I:2733, 1988.
16. R. Mech and P. Prusinkiewicz. Visual models of plants interacting with their
environment. in proceedings of siggraph96.ACM Press, 30(4):397410, 1996.
17. P. Prusinkiewicz, J. Hanan, M. Hammel, , and R. Mech. Lsystems: from the theory
to visual models of plants.Machine Graphics and Vision, 2(4)(1508):1222, 1993.
18. O. Deussen, P. Hanrahan, B. Lintermann, R. Mech, M. Pharr, and
P. Prusinkiewicz. Realistic modeling and rendering of plant ecosystems. in proceed-
ings of siggraph98, annual conference series 1998.ACM Press, page 275286, 1998.
19. Bedrich Benes. and Juan Miguel Soto Guerrero. Clustering in virtual plant
ecosystems.WSCG SHORT Communication papers proceedings., 2004.
20. M. Alsweis and O. Deussen. Modeling and visualisation of symmetric and asym-
metric plant competition.Eurographics Workshop on Natural Phenomena., 2005.
21. O. Deussen and B. Lintermann. A modelling method and user interface for
creating plants. InProc. Graphics Interface 97, pages 189–198. Morgan Kaufmann
Publishers, May 1997.

Multi-layered Stack Mosaic
with Rotatable Objects
Jin Wan Park
1
, Kyung Hyun Yoon
2
, and Seung Taek Ryoo
3
1
GSAIM Dep. Chung-Ang Univ.
221 Heukseok-dong, Dongjak-gu, Seoul, Korea
[email protected]
2
CS Dep. Chung-Ang Univ.
221 Heukseok-dong, Dongjak-gu, Seoul, Korea
[email protected]
3
Software Dep. Han-Shin Univ.
411 Yongsan-dong, Osan-si, Gyonggido, Korea
[email protected]
Abstract.We present a new type of mosaic, the multi-layered stack
mosaic with photographs of rotated objects. Our algorithm uses multi-
layered Photomosaics with database enrichment by element rotation. The
beneﬁt of this algorithm is that an artist can not only produce a digital
mosaic with a relatively small database without degrading the quality
of the mosaic, but that the Multi-layered Stack Mosaic also generates
unique and strong artistic expressions which gives an illusion of piled
stackable objects. Since the result has a unique visual style, we intend to
exhibit our mosaic images at galleries such as the Epson Color Imaging
Contest and the CAU Art Center
1
.
1 Introduction
The mosaic is an intriguing subject in computer graphics because it involves sci-
entiﬁc and artistic research simultaneously. The mechanism of traditional mosaic
creation is based on the visual perception of artists. Although it is diﬃcult to
quantify, the process is systematic and methodical. In recent years, many mosaic
schemes for generating free-form 2D images have been suggested in the area of
non-photorealistic rendering. In 1996 Robert Silvers devised the concept of Pho-
tomosaics [1] in which one big picture is generated from thousands of smaller
pictures. These pictures are then placed in patterns like traditional mosaics, and
each cell is an actual photograph. Each tile of a Photomosaic contains a photo-
graphic image, causing perceptual confusion in the observer in a delightful way.
The perceptual range changes from elements to the whole and vice versa. How-
ever, for producing quality Photomosaic artwork, an abundant image database
must be prepared making this one of the most laborious aspects of Photomo-
saics. What makes this worse is that every picture may have its own copyright
1
This work was supported by grant No. (R01-2005-000-10940-0) from the Basic Re-
search Program of the Korea Science & Engineering Foundation.
H.-P. Seidel, T. Nishita, and Q. Peng (Eds.): CGI 2006, LNCS 4035, pp. 12–23, 2006.
cﬀSpringer-Verlag Berlin Heidelberg 2006

Multi-layered Stack Mosaic with Rotatable Objects 13
license. We will present a new type of mosaic scheme which contains elements
of stacked and rotated objects. Although our algorithm is an outcome of scien-
tiﬁc research, we also focus on the artistic side of mosaics with regard to visual
expression. In other words, our goal is not only producing quality mosaics with a
limited picture database size, but also creating beautiful and impressive mosaics
as art. This paper is organized as follows: related research of mosaics are pre-
sented in the next section; in section 3, we analyze and evaluate our algorithm;
in section 4, resulting images will be presented.
2 Related Research and Issues
As we can judge from the thousands of years of history, the mosaic is one of the
oldest methods of artistic expression. Various mosaics, in which a large image is
formed by a collection of small images can be created depending on the type of
tiles used and the restrictions in their placement [2][3][4]. In the case of each tile
as meaningful subject, for example, Guiseppe Arcimboldo [5] in 16th century,
and Chuck Close in 20th century [6] have the common feature of multi-range
perceptive target sizes. This type of mosaic is revised again by Robert Silvers in
his Photomosaics, a collection of small images that are arranged in a rectangular
grid and form a larger image in the distance. This has strong expression because
there is no photograph that is absolutely free from the photographer’s opinion.
When we use photographs as elements of mosaics, the mixed perceptual range
provides pleasurable games and challenges for the observer. A rich photograph
database is the essential condition to produce good Photomosaics. This huge
database is not only hard to build, but also extremely expensive to use due
to the fact that each picture may carry a copyright license. There are some
well known tricks to reduce the number of pictures in the library, such as self
duplication by ﬂipping or resizing images, or by changing color value, but the
damaged images can degrade the quality of Photomosaics because each element
is as an important perceptual subject in relation to the whole image. We seek
to prevent the loss of the each picture’s signiﬁcance.
3MethodOverview
The method we present in this paper is divided into two primary steps.
–Rotate the object to improve the possibility of a better match. The meaning
of each photograph has to be preserved and garbage data due to rotation
has to be minimized.
–Fill the empty space (hole) that is produced in the above step, with multiple
layers.
Each step is important to visualize the unique style of the suggested mosaic.
This also improves the likeness of two images, a source image and an output
mosaic image, so we can get a better output result compared to the conventional
Photomosaics algorithm under the same database conditions especially when the
database size is small.

14 J.W.Park,K.H.Yoon,andS.T.Ryoo
3.1 Objects Rotation
Image Selection Process.The Photomosaic algorithm is used for image selec-
tion and database management[10]. It does not attempt shape or edge analysis.
Rather, it seeks to minimize a distance measure based purely on diﬀerences of
pixel values that are its red, green, and blue components [12]. The original im-
age is divided into rectangular blocks of the same dimensions as the tiles. The
algorithm scans the blocks of the original images from left to right and top to
bottom, replacing each block with a tile before considering the next. Thus, each
pixel in each cell of the original image is compared with each pixel of every
picture in the database. However, it takes a huge amount of time to search for
the best image [10][11]. As suggested by Silver, to improve the speed we use an
index system that reduces candidate images. With this Photomosaic algorithm,
each block of the original image will be replaced with the best matching; in other
words the most similar photograph in the database. There is no color correction
in this process.
A
w×his the image’s rectangular blocks as a matrix whileT
k
w×h
is tile k as a
matrix, andA
r(i, j),A g(i, j),A b(i, j) are the red, green and blue components
of a given pixel. The Photomosaic algorithm seeks to minimize the following
distance over all tiles in the library.
d=min
k
⎡
⎣
w

i=1
h

j=1

A
r(i, j)−T
k
r
(i, j)

+

A g(i, j)−T
k
g
(i, j)

+

A b(i, j)−T
k
b
(i, j)

⎤
⎦
(1)
To increase the variety measure of the generated Photomosaics, two modiﬁ-
cations are made to the basic algorithm. First, a limit is imposed on the number
of times a tile can be selected. Second, a minimum distance is required between
any two copies of a tile. To implement these changes, the count and last selected
position is maintained for each tile [10].
Rotation Sampling Rate.The traditional Photomosaic algorithm does not
alter the database so the quality of the output image depends on the number
of collected images. We suggest that without collecting extra images we can im-
prove the result by enriching the library; by rotating photographs in the library.
Theoretically we could reproduce inﬁnite copies of each image with this trick,
but the fact is that rotated images are very similar with each other, so the ro-
tation does not eﬀectively enrich the library when the angle is close between
images. Note that more pictures are not always better because this will slow
down the database searching time. Therefore, it is important to ﬁnd a balance
between quality and price although the best case varies depending on a type of
project. It is clear that a dense sampling rate will guarantee the best result, but
the degree of improvement is not linear. We show the relation between sampling
rate and similarity of two images in this section. As we can see the follow-
ing test, the steep improvement becomes negligable with ﬁner samples. To test
under the harshest environment, we limit the picture database to one hundred

Multi-layered Stack Mosaic with Rotatable Objects 15
Fig. 1.Part of the coins that are used for artworks. The small coins and big coins are
mixed to reduce patterns.
Fig. 2.2PI/N Sampling (N=1 Left), (N=128 Right) the ﬁner sampling, the better
output
coins produced by the United States mint oﬃce. The pictures do not have to be
circular in shape, but this shows the features of our algorithm more easily.
Depending on the type of images used in the test, the number in the table
will be changed but the rate of increase remain constant. Average color distance
means the average distance of each corresponding pixel between the original
image and the output image in green, red and blue components
Suitable Object for Rotation.Rotating photographs generates problems.
Traditional Photomosaics place a photograph in a rectangular cell. This naturally
matches with rectangular shaped pictures in the database. When we rotated
pictures, two main problems appear. First, as we can see from Figure 3, it spoils
the contents of a photograph, and second, it generates unwanted data in the cell.
The unwanted data has two types, ﬁrst is garbage data or empty space (hole)
which is the red part, ’A’, and the other is the data loss part truncation which
is the blue area, ’B’, in the Figure 3.
Most pictures are intolerant to rotation because they have a deﬁnite orienta-
tion. When an image is rotated, it becomes an obstacle to perceive the picture
as a meaningful object. The conditions for a safe image, when it is rotated, are
as follows (Figure 4).

16 J.W.Park,K.H.Yoon,andS.T.Ryoo
Fig. 3.Rotation of a picture destroys the meaning of the photograph. Image rotation
generates unwanted data in the cell-’A’ and data loss-’B’.
Fig. 4.Rotation of a coin image. Sampling rate is 30 degree. It produces 11 more
images from the original image.
Fig. 5.Left images show examples of Tolerant Objects by rotation. Right image shows
data loss by rotation. The object should be in the inscribed circle area in the cell to
prevent any data loss by truncation.
The ﬁrst type of images that are rotation invariant are of objects that are
viewed from above. (Figure 5 Left). The second type are ones that have less
data loss and produce less garbage data by rotation. Valid data is contained in
a circle inscribed in the cell (Figure 5 Right). As we can see from Figure 4, the
object ’Coin’ satisﬁes these conditions.
When the cell is a square and an object in a photograph is a circular shape,
this produces the least unwanted data because the two conditions maximize the
trustable area; the inscribe circle area is the only trustable data.
3.2 Stacking Objects – Multilayering with Angled Grid System
Angled Grid System.The method in the previous section inevitably produces
garbage data. The inscribed circular area is the region we may trust in a square

Multi-layered Stack Mosaic with Rotatable Objects 17
Fig. 6.Single layer. White area is the ’Empty space’(left). Double layers , less ’empty
space’ than the single layer(center). Multilayer eliminates the empty space from top to
bottom, 1 layer, 2 layers, and 3 layers stacked (right).
cell. The other area in the cell, we deﬁne as empty space (hole
has no information. Due to the law of simplicity in Gestalt theory [7] we can still
recognize the larger image without trouble. However, there is a way to improve
the quality of mosaic when we use the empty space aggressively by replacing
the empty space with the information from other layers [8]. The angled grid
system was designed for this process. The blocks in traditional Photomosaics
are aligned along horizontal and vertical lines, and this is suitable because the
photograph in each block is rectangular in shape and should be aligned upright
to be recognized. This makes output images uniform (e.g Figure 6). The angled
grid system has two advantages. First, it eliminates empty space eﬃciently when
the layers are stacked. Second, it produces fewer patterns. The degree of grid
rotation is free as long as it does not produce signiﬁcant patterns while covering
empty space well. As we can see from Figure 6, multi-layering in various angles
not only covers the empty space but also produces a strong illusion of randomly
stacked objects.
Multiple Layers. How Many?Using multiple layers has two important roles.
First it improves the quality of the mosaic, and second, it gives the eﬀect of piling
objects. As we can see from the Figure 6, multilayering covers the empty space
eﬀectively. The question is, how much it will improve the quality when another
layer is added?
The right of Table 1 shows a similar result to what we saw in the section
3.1. More layers produce better output, but too many layers cost too much
rendering time while only improving output quality slightly. More than three
layers do not help the quality much, so in general we use three layers to produce
the artworks shown. When we deﬁne x as the similarity between the result and
the original image, we can expect that single layer mosaics describe a similarity
as (PI/4×x) because the inscribe circle area is the trustable area. When we
apply the multilayer mosaic method, the empty space of each cell is eliminated
and we can expect a better result than the single layer. The multilayer mosaic
without the empty space is 4/P Itimes better than the single layer mosaic. In
other word, we can expect a similarity to traditional Photomosaics without using
rectangular, uniformed tiles.

18 J.W.Park,K.H.Yoon,andS.T.Ryoo
Table 1.Average color distance value of pixels with each rotation sampling rate(left).
The more layers, the better quality(right).
Sampling RateAvg. distanceNumber of layersAvg. distance
2PI/1 34.994 1 29.673
2PI/2 32.466 2 27.318
2PI/4 31.659 3 26.066
2PI/8 31.018 4 25.535
··· ··· 5 25.206
2PI/64 30.105 ··· ···
2PI/128 30.103 ··· ···
3.3 Shadow
The multilayer mosaic generates the spatial order in elements. Applying a shadow
eﬀects to them creates a stronger illusion of depth. For shadow eﬀects, we treat
each layer as a two dimensional thin ﬁlm rather than a three dimensional object.
We simply spread out the edge of each object and darken the edge area. This
simple process creates a very strong sense of depth.
Fig. 7.Before adding shadows to each layer vs. After adding shadows to each layer
4Result
4.1 Circular Objects – e.g. Coins
Figure 9 and 10 are two artworks made using the suggested algorithm. Two
hundred coin images are used, and the rotation sampling rates are ten degrees.
We used three layers for these images. These artworks received an award from the
Epson Color Imaging Contest, a competitive international contest in traditional
photography. This suggests the judges were convinced that these images are real
photographs. It was a unique experience because the output has both artistic
and technical issues. To verify the quality of the algorithm and to prove the
artistic value of the images, we showed the results to public to be judged by
experts trained in the arts.

Multi-layered Stack Mosaic with Rotatable Objects 19
Fig. 8.B,C,D shows empty space as red color
4.2 Non-circular Objects – e.g. Human Body
As we mentioned in Section 3.1.3 the objects in our library do not have to be
circular in shape as long as the valid data is inside of inscribed circle area (Figure
8-A). A circular object has the biggest valid data area, but it does not guarantee
the minimum color diﬀerence between original source image and output mosaic
image. The most important part of this algorithm is that we do not want to solve
all the problems in a single layer because the other layers will help to complete
the mosaic. The more layers, the less empty space that remains. Figure 8 shows a
non-rectangular shaped object. In the case of coins, we deﬁned the empty space
as non-inscribed circle area (Figure 8-B), but for these pictures we deﬁne the
empty space as the non-object area in red (Figure 8-C,D).
Figure 11 is an example mosaic produced by the suggested algorithm. We used
three layers and a library with eighty pictures. None of them is a circular-type
object, so we blue-screened the background when we shot the ﬁgure of a female
dancer. This dancer mosaic was selected for the Shichon-Seoul Art Festival 2004
main poster.
5 Conclusion and Future Work
The Multi-layered stack mosaic algorithm with photograph rotation improves
image quality, especially when the database is limited. We also tested this new
mosaic style by opening it to the public with a gallery exhibition [9]. The exhi-
bition was successful and the images were well received. However, some people
complained that they are too mechanical. Therefore, in future work, we hope to

20 J.W.Park,K.H.Yoon,andS.T.Ryoo
Fig. 9.Whisky. 200 coins of database.

Multi-layered Stack Mosaic with Rotatable Objects 21
Fig. 10.Mosaice of a girl with a bag. 200 coins of database.

22 J.W.Park,K.H.Yoon,andS.T.Ryoo
Fig. 11.As we can see detailed ’A’ area, pile of small pictures of a dancer makes
smooth curve in bigger mosaic picture.

Multi-layered Stack Mosaic with Rotatable Objects 23
reﬁne this algorithm with a ﬂexible grid to create a more natural looking collage.
This will be a challenge because the algorithm has to understand the substance
of each object. The study of how artists select objects for painting is essential
to move on to the next step of art in science.
References
1. Silver, R and Hawley, M.: Photomosaics. New York, Henry Holt (1997)
2. Hausner A, ”Simulating decorative Mosaics”, In: Proceedings of ACM SIGGRAPH
2001: 573-580
3. Kim J, Pellacini F, ”Jigsaw Image Mosaics”, In Proceedings of ACM SIGGRAPH
2002,: 657-664
4. Gershon Elber, George Wolberg, ”Rendering traditional mosaics” The Visual Com-
puter 2003: 67-78.
5. Strand, C. Hello, Fruit Face! The Paintings of Guiseppe Arcimboldo, Prestel (1999)
6. Thomas G. West, Artist discoveries and graphical histories, ACM SIGGRAPH
Computer Graphics, Volume 33 , Issue 4 Nov.2000 12-13
7. Koﬀa, K. : Principles of Gestalt Psychology, London, Routledge and Kegan Paul
(1935)
8. Rensink, R. A. : Seeing, Sensing, and Scrutinizing. Vision Research (2000) 40,
1469-1487
9. Park, Jin Wan : The Coinage Project No.1 ”Time Is Money”, Leonardo 38:2, MIT
Press (2005)
10. Nicholas Tran: Generating Photomosaics: An Empirical Study. SAC1999:105-109
11. Tom McKenna and Gonzalo R. Arce. New Image Mosaic Structures. Technical re-
port, De-partment of Electrical and Computer Engineering, University of Delaware,
August 1999.
12. Adam Finkelstein and Marisa Range. Image Mosaics. Technical Report TR-574-98,
Princeton University, Computer Science Department, March 1998.

Appearance and Geometry Completion with
Constrained Texture Synthesis
Chunxia Xiao, Wenting Zheng, Yongwei Miao, Yong Zhao, and Qunsheng Peng
State Key Lab of CAD&CG, Zhejiang University, 310027, Hanghou, P.R. China
Abstract.A novel approach for appearance and geometry completion
over point-sampled geometry is presented in this paper. Based on the
result of surface clustering and the given texture sample, we deﬁne a
global texture energy function on the point set surface for direct texture
synthesis. The color texture completion is performed by minimizing a
constrained global energy using the existing texture on the surface as
the input texture sample. We convert the issue of context-based geom-
etry completion into a task of texture completion on the surface. The
geometric detail is then peeled and converted into a piece of signed gray-
scale texture on the base surface of the point set surface. We ﬁll the holes
on the base surface by smoothed extrapolation and the geometric details
over these patches are reconstructed by a process of gray-scale texture
completion. Experiments show that our method is ﬂexible, eﬃcient and
easy to implement. It provides a practical texture synthesis and geometry
completion tool for 3D point set surfaces.
1 Introduction
As numerous 3D surface scanning devicesare available in recent years, 3D scan-
ning has become a major approach for acquiring the shape of complex 3D objects.
However, obtaining a ﬁne and usable 3D model from the acquired surface sam-
ples is still a diﬃcult task. Due to occlusions, low reﬂectance, measure error in
the scanning, the acquired geometry is frequently imperfect, that is, it contains
holes. In addition, large holes may also be introduced by some surface editing
operations. These holes have to be ﬁlled in a manner not only conforming to
the global shape of the entire surface but also exhibiting its primary geometric
detail. At the same time the color textureof the defective surface should also be
repaired consistently.
Compared with the problem of inpainting and texture synthesis of 2D image,
the geometry completion and texture synthesis on 3D point-sampled geometry is
more challenging for many reasons. The point sampling is irregular and does not
give rise to a regular parameter domain as the image. In addition, the similarity
measurements between the point sets are diﬃcult to deﬁne. In this paper, we pro-
pose a novel method for appearance and geometry completion on point-sampled
surface, our major contributions are as follows:
Firstly, Based on global optimization we present a novel texture synthesis
producing smooth texture synthesis eﬀects while keeping the intrinsic structures
H.-P. Seidel, T. Nishita, and Q. Peng (Eds.): CGI 2006, LNCS 4035, pp. 24–35, 2006.
cﬃSpringer-Verlag Berlin Heidelberg 2006

Appearance and Geometry Completion with Constrained Texture Synthesis 25
of the sample texture over the point-sampled geometry. By regarding the tex-
tures on the existing surface as the input texture sample, the texture completion
can be accomplished by optimizing a constrained global texture energy func-
tion. Secondly, By performing a mean curvature ﬂow we derive the base surface
of the sampled geometry. We deﬁne the geometric details as the displacement
between the surface and its base surface. The geometric details are then peeled
and converted into a signed gray scale texture attached to the base surface for
the downstream processing. Thirdly, our approach reconstructs the geometric
details on the smooth patch by implementing texture completion in the gray
scale texture space, therefore some troublesome operations such as similarity
measurement, rigid transformation of the 3D points set are avoided.
By applying the above algorithms we can achieve consistent context-based
completion and can deal with more complex boundary condition compared with
the texture and geometry completion algorithms employing PDE. Furthermore,
by converting the 3D geometry completion into the task of 3D texture comple-
tion, we are able to further utilize a wealth of currently available surface texture
synthesis and completion technique to serve our purpose.
2 Related Works
There has been a number of works focused on Example-based texture synthesis
for 3D surface. A comprehensive survey isoutside the scope of this paper. It is
found that most of existing texture synthesis algorithms on 3D surface are mesh
oriented. Since texture synthesis algorithms on meshes normally make use of the
topology information, they cannot be applied to the point set surface directly.
By now, few works are focused on texture synthesis over point set surface. In
our scope, Alexa et al. [1] and Clarenz etal. [2] showed some texture synthesis
results on point set surface as the application of their algorithm. Recently, geom-
etry completion focused on repairing the uncompleted meshes or point-sampled
surface has received much attention in computer graphics. Lots of works have
been presented, and these methods can be mainly divided into two categories
according to the strategies they adopted.
One strategy is to create a smooth patch covering the hole-region and satis-
fying the boundary conditions. [3] used globally supported radial base functions
(RBFs
a hierarchical approach to 3D scattered data interpolation with compact RBFs.
Davis et al. [5] constructed a volumetric signed distance function around the sur-
face samples, and then applied an iterative Gaussian convolution to propagate
adjacent distance values to patch the holes. Liepa [6] proposed a hole ﬁlling tech-
nique to interpolate the shape and density of the surrounding mesh. Verdera et al.
[7] extended a PDE-based image inpainting technique to 3D geometry meshes.
Ju [8] constructed an inside/outside volume using octree grids for any model
represented as a polygon soup, then the holes were repaired by contouring.
The other strategy is to repair the holesaccording to the context information
so that the geometry detail can be reconstructed at the same time. Sharf et al. [9]

26 C. Xiao et al.
introduced a context-based method which extended the texture synthesis tech-
niques from 2D image to 3D point-based models for completing the defective
geometry models. Pauly [10] presented a method using a database of 3D shapes
to provide geometric priors for regions of missing data. Lai et al. [11] proposed
a method of geometric detail synthesis and transferring for meshes based on the
concept of geometry images. Park et al. [12] restored both shape and appear-
ance from the incomplete point surfaces by conducting a local parameterization
to align patches and then solved a Poisson equations on 2D domain for warping
the patch to cover the hole region. Minh et al. [13] transformed the 3D geome-
try synthesis problem into the 2D domain by parameterizing surfaces, and then
solved the geometry completion problem with an interactive PDE solver.
3 Texture Synthesis
Recently, Kwatra et.al [14] presented an approach for 2D texture synthesis based
on global optimization of texture quality with respect to a similarity metric
based on Markov Random Field (MRF
extend this method to point-sampled surface. As the point surface is irregularly
sampled in 3D, it is diﬃcult to deﬁne such a kind of global texture energy on
point-sampled geometry.Preprocessing of the pointset surface is necessary as
the preparation.
The point-sampled geometryM={p1,p2, ...pn}is ﬁrstly clustered into uni-
form patches{Co,i}that are the units for further computing. The neighboring
patches are overlapped to make it less computationally expensive for comput-
ing the energy, furthermore, avoid the synthesized texture getting blurred in
regions where there is a mismatch between the overlapping clusters. We then set
up a global continuous direction ﬁeld on the point set surface to conduct local
parameterization for each patch. By building up the correspondence between
irregularly 3D sampling points and the regular 2D texture samples, the global
texture energy can be deﬁned directly on the surface and to be optimized.
3.1 Surface Clustering
We ﬁrst utilize the hierarchical clustering algorithm [15] to split the point cloud.
The point cloudMis then divided into a number of subsets{C
ﬀ
i
}. Nevertheless,
these initial clusters contain sharp edges and corners, as shown in Fig.1(a). Let
y
ﬃ
i
be the centroid of the clusterC
ﬀ
i
. To get a more even distribution of clusters,
we ﬁnd neighborNi={j:0<=ﬃy
ﬃ
j
−y
ﬃ
i
ﬃ<r}for each pointy
ﬃ
i
.Foreach
point samplepi∈C
ﬀ
i
,welocatey
ﬃ
j
∈Nithat is the nearest topi,andthenpiis
adjusted to the new clusterCj. Therefore, the partitioning result of the entire
point set is reformed to uniform clusters{Ci}, as illustrated in Fig 1(b).We then
grow each{Ci}and form the new generated clusters{CO,i}, such that eachCO,i
overlaps with its neighboring clusterCO,j. Within a band of widthh.Inour
experimentation we seth=0.5·d,wheredis the average radius of the clusters
{Ci}. In Fig1. (c tes the overlapped area of adjacent
clusters, in Fig1. (d htakes a larger value.

Appearance and Geometry Completion with Constrained Texture Synthesis 27
(a) (b (c (d (e
Fig. 1.Surface clustering of point-sampled geometry. (a) Initial clustering, (b) uni-
form clustering, (c) overlapped cluster with small overlapping parameter, (d
overlapping area, (e) The texture synthesis result.
3.2 Global Optimized Surface Texture Synthesis
We establish the texture energy function directly on the point set surface in
several steps. We ﬁrst deﬁne the global energy on the point-sampled surface.
Letyibe the centroid of the clusterCO,i. By applying the method presented
in [1], we can establish a global direction ﬁeld on the point set surfaceY=
{y1, ...ym}. Based on the direction ﬁeld and the tangent plane computed by
covariance analysis, we set up a local frame for each clusterCO,ito facilitate the
local parameterization.
Letνibe the direction ofyi,andnibe the normal ofyi. We deﬁneνithe
up direction andμi=νi×nithe right direction for the texture, as shown in
Fig. 2(a). A local frame{μi,νi,ni}is then established.We project the vectors
fromyito all surrounding pointsxiin the cluster∈Co,ionto the tangent plane of
yi. The resulting vectors are normalized and multiplied by the distance between
yiandpi(Fig. 2(a)). In this way we preserve the distance information between
the points on the surface to some extent.
(a) (b (c (d
Fig. 2.the construction of the grid for each patch (cluster). (a) Local parameterization
by projection, (b) regular grid G generated on the parameterization domain, (c) to
generate the sample texture from user speciﬁed region. (d) mean curvature ﬂow.
Now a regular parameterization gridGiofn×ncentered onyiwith the interval
ofhon the tangent plane is generated for eachCO,i,Gialigns with the direction
νias up vector and theμias right vector (Fig.2(b)). The parameter at each grid

28 C. Xiao et al.
(a) (b (c (d
Fig. 3.Some texture synthesis results based on the global optimization
pointtijis obtained by interpolating the parameters of points in the surrounding
cells. So with the regular gridGi, the correspondence between irregularly 3D
sampling points and the regular 2D texture samples can be built for deﬁning the
global texture energy. LetX={Gi}denote the surface parametric space over
which we want to compute the texture energy andZdenote the input texture
sample. LetZibe the vectorized pixel neighborhood inZwhose appearance
is most similar toGiunder the Euclidean norm. Then, we deﬁne the texture
energy overXto be
Et(x;{Zi})=
ﬃ
p∈Y
ﬃGi−Ziﬃ
2
(1)
Similar to the algorithm presented in [14], we use the EM-like algorithm to
optimize the texture energy overX.WemodifytheEandMsteps to account
for the speciﬁc feature of the discrete point set.
TheM-step of our algorithm minimizes (1
put neighborhoods{Zi}, keepingGiﬁxed and for eachGi, we ﬁnd its nearest
neighborZifromZ.
In theE-step, we need to minimize (1) w.r.t.Gi. We resolve it in a way diﬀer-
ent from [14]. Once we ﬁnd the closest input neighborhoodZifor eachGi,the
texture intensity at pointpienclosed inGican be obtained using bilinear inter-
polation. Suppose clusterCO,iandCO,joverlap with each other, some points
will be contained in bothCO,iandCO,j.Eachofthecommonpointsmaytake
possibly diﬀerent intensity values fromGiandGj. The minimization procedure
assigns each common point an intensity value that is equal to the average of the
original values inCO,iandCO,j. Since the intensity at the common point has
changed, the intensity attijinGiis updated which is used for the nextM-step.
The energy of the synthesized texture will converge after a number of iter-
ations. Fig.1 (e
Multi-scale synthesis can also be performed in our approach by adjusting the
gridGi(n×n).
4 Color Texture Completion
In this section, we complete the color texture of the point-sampled geometry
with a constrained texture synthesis algorithm, and the completed color texture

Appearance and Geometry Completion with Constrained Texture Synthesis 29
should be consistent with the surrounding existing texture. We select a user
speciﬁed regionDon the existing surface to serve as input sample texture. For
each pointpiinD, similar to the method presented in section 3.3, we ﬁnd its
neighborhoodNiand build a colored regular GridZiofn×n,asshownin
(Fig.2(c {Zi}is used as the input texture for color completion.
To make the boundary between the completed and original regions impercep-
tible, similar to the controllable image synthesis [14], we add a additional term to
Eq.(1) to achieve a general constrained energy function for texture completion.
Et(x;{Zi})=
ﬃ
p∈Y
ﬃGi−Ziﬃ
2
+
ﬃ
k∈ϕ
(x(k)−x
c
(k))
2
(2)
In our approach,ϕis the set of boundary points, or the set of boundary clusters
which contain some boundary points,x
c
is a vector containing the current color
values at the boundary points.
(a) (b (c (d
Fig. 4.Texture completion. (a)Original point model,(b)defective texture model, (c)the
defective regions is covered by the clusters,(d)completed texture point model.
Note that to generate a seamless boundary texture, the cluster set{CO,i}
should be chosen to cover the color defective regions and contain the boundary
points, as shown in Fig.4(c).As shown in Fig.4(b), the region enclosed by the
blue curve is used for generating the sample texture, the red patch is the re-
gion to be completed, the result is shown in Fig.4(d). In Fig.5, we present the
completion result with isolated islands left on the ﬂawed region. With our con-
strained texture ﬁlling technique, the result is consistent with islands as well as
the boundary of the existing surface color. This situation is diﬃcult to handle if
one applies the Poisson equation interpolation method [12].
5 Geometry Detail Encoding
The geometric detail is an important attribute of a surface. It is deﬁned as the
diﬀerence between the original point-sampled surface and its base surface. In our
method, the base surface is built by smoothing the point set surface. Covariance
analysis on local point cloud can be applied to estimate various local surface
properties, for example, normalν0and the curvatureσnat each pointpion the
point set surface[15]. Based on the normal and curvature, we are able to deﬁne a
curvature ﬂow equation[16]. The basic idea of deﬁning such a diﬀusion ﬂow is to
allow the point moving along the normal with a speed equal to the curvatureσn

30 C. Xiao et al.
(a) (b (c
Fig. 5.(a) The original Chameleon model, (b) corrupted texture model with islands,
(c
as shown in Fig.2 (d). Since the position of a point is adjusted along the normal
direction, the proposed curvature ﬂow will not introduce undesirable point drift-
ing over the surface. The curvature ﬂow equation is an isotropic smoother and
a low frequency base surface can be received applying this ﬁltering operator.
5.1 Encoding Geometric Details as Texture
Assume that the surfaceMis ﬁltered, yielding a base surfaceM
ﬀ
.Letp∈M,p
ﬀ
∈
M
ﬀ
is its corresponding point, andn
ﬀ
is the normal at pointp
ﬀ
.Letδ=ﬃp−p
ﬀ
ﬃ
be the geometric detail of pointp.Letdire=(p−p
ﬀ
)·n
ﬀ
,ifDire≥0, then
sign=1,elsesign=−1. We deﬁnec
ﬀ
=sign?ﬃp−p
ﬀ
ﬃas the the signed
gray-scale ofp
ﬀ
.
Once the normaln
ﬀ
andsignedgray-scalec
ﬀ
ofp
ﬀ
are obtained, its geometry
information is approximately reconstructed as p=p
ﬃ
+c
ﬀ
·n
ﬀ
. Using this technique,
asurfaceMis reconstructed by the base surface and the signed gray-scaleC
which approximatesM. The normal information of the reconstructed points can
be recomputed using the minimum spanning tree [17].With the mean curvature
ﬂow ﬁlter, the reconstructed surface from the signed geometric gray level is a
good approximation to the original surface, as shown in Fig.7. With diﬀerent
iteration times for the mean curvature ﬂow, various frequency band of geometric
detail can be extracted eﬀectively and eﬃciently.
6 Geometry Completion
Similar to the color texture completion, the completed geometry should keep
consistent with the surrounding geometry and the boundary between the com-
pleted and existing regions should be continuous. Using a hierarchical compactly
supported basis functions [4], we ﬁrst complete the base surface by smoothed
extrapolation, as shown in Fig.6(d
surface patch should be reconstructed.
6.1 Context-Based Geometry Completion
Using the method presented in previous sections we can complete the signed
gray-scale texture on the patched smooth surface (Fig.6 (e

Appearance and Geometry Completion with Constrained Texture Synthesis 31
(a) (b (c (d
(e (f) (g) (h
Fig. 6.Overview of our entire processing pipeline for geometry completion. (a) The
defective bunny model, (b) the base surface of (a), (c) the signed gray-scale texture on
the base surface. Pseudo color is used to illustrate the value clearly. Red color indicates
the largest gray-scale value, while the blue indicate the least value, (d) completed base
surface, (e) the completed signed gray-scale texture in pseudo color, (f) the ﬁnal result
of geometry completion, (g) RBF interpolation of (a), (h
signed gray-scale texture is then converted back to geometric details (Fig.6 (f)),
and the context-based geometry completion is achieved. This procedure is re-
garded as a reverse procedure of the geometric detail encoding.
Letp∈Mwith normaln,p
ﬀ
is its corresponding point onM
ﬀ
with normaln
ﬀ
.
We interpolate the defective base surfaceM
ﬀ
to get the completed base surface
N
ﬀ
. SupposeΩ
ﬀ
=N
ﬀ
−M
ﬀ
is the newly constructed patch which is consistent
with the boundary of theM
ﬀ
, we then complete the textureC
ﬀ
ofΩ
ﬀ
based on
the existing texture onMemploying the technique presented in section 4. For
each pointν
ﬀ
∈Ω
ﬀ
with normaln
ﬀ
,letc
ﬀ
be its synthesized signed gray-scale, the
reconstructed location is deﬁned asν=ν
ﬀ
+c
ﬀ
·n
ﬀ
, its normal can be recalculated
using the technique described in [17]. Using this approach, the completed patchΩ
captures the context information of theexisting surface. Compared with other
geometry completion approach [9] that added points by rotating, translating,
and possible warped copies of points from another region, our approach is more
eﬃcient, controllable and easy to implement.
AssumeNbe the reconstructed surface based onN
ﬀ
and its signed gray-scale.
Ω,MandNare all continuous inside their interior region. When we adopt the

32 C. Xiao et al.
(a) (b (c (d
Fig. 7.Surface reconstruction from the signed gray scale texture. (a) Original Bunny
modelM, (b) Smoothed Bunny modelM
ﬃ
, (c) Signed gray-scale texture in pseudo
colorC, (d) Reconstructed surface M.
points fromMand the other points fromΩin the ﬁnal completion result, small
crack may occur in the boundary region betweenΩandM.Toresolvethis
problem, we modify the normal of the boundary points inΩ
ﬀ
.
Supposepis a boundary point onM,andp
ﬀ
is its corresponding point on base
surfaceM
ﬀ
with normaln
ﬀ
=(n
p
ﬃ
x
,n
p
ﬃ
y
,n
p
ﬃ
z
). Letδ=(δx,δy,δz) be the normalized
vector of the diﬀerence vectorp−p
ﬀ
. The vector diﬀerenceΔμp
ﬃbetweenδand
n
ﬀ
is deﬁned asΔμp
ﬃﬃ(Δ
p
ﬃ
x
,Δ
p
ﬃ
y
,Δ
p
ﬃ
z
)ﬃ(δx−n
p
ﬃ
x
,δy−n
p
ﬃ
y
,δz−n
p
ﬃ
z
).
We deﬁne a band with radiusdaround the boundary of the smoothed surface
patchΩ
ﬀ
.Foreachpointq
ﬀ
=(xq
ﬃ,yq
ﬃ,zq
ﬃ)withnormalnq
ﬃ=(n
q
ﬃ
x
,n
q
ﬃ
y
,n
q
ﬃ
z
)in
the band, we ﬁnd its nearest pointp
ﬀ
on the boundary ofM
ﬀ
. Let the distance
betweenp
ﬀ
andq
ﬀ
bes,wesettheweightω=(d−s)/d, then the normalnq
ﬃof
q
ﬀ
is adjusted as (n
q
ﬃ
x,n
q
ﬃ
y,n
q
ﬃ
z)ﬃ(n
q
ﬃ
x+ω·Δμ
p
ﬃ
x,n
q
ﬃ
y+ω·Δμ
p
ﬃ
y,n
q
ﬃ
z+ω·Δμ
p
ﬃ
z).
By normalizing above vector we get its modiﬁed normaln, then the geometry
position ofq
ﬀ
can be reconstructed asq=q
ﬀ
+c
ﬀ
·n. As the completed base surface
N
ﬀ
is continuous, its normal is also continuous. In addition, the completed texture
Cis continuous around the boundary region. By using this technique, we are
able to produce a continuous surface around the boundary and the holes are
ﬁlled consistent to the existing surface(Fig.6(f)).
6.2 Geometry Completion with Structure Propagation
The synthesis order of geometry completion is also important. By augmenting
texture synthesis with some automatic guidance or interactive guidance, it can
signiﬁcantly improves the quality of completion by preserving some salient struc-
tures. Similar to Sun et. al. [18], we propose a geometry completion method based
on structure guided synthesis. The missing structure information is speciﬁed by
extending a few curves or line segments from the known regions to the unknown
regions, then the patches along these user-speciﬁed curves in the unknown re-
gion is synthesized using patches selected around the curves in the known region
by using a global optimization. After the salient structure is completed, the
remaining regions can be completed. In our method both the salient structures
and remaining missing regions are completed based on the constrained global

Appearance and Geometry Completion with Constrained Texture Synthesis 33
(a) (b (c (d
(e (f) (g) (h
Fig. 8.Geometry completion with structure propagation. (a) Original Venus model,
(b M, (c) the base surfaceM
ﬃ
ofM, (d) the geometry detail of
Mis converted to the signed gray-scale texture onM
ﬃ
, (e) One red line speciﬁed by
the user in completed base surface, the yellow region is the structure region needed
be synthesized ﬁrstly. (f) intermediate result after synthesizing structure and texture
information along the user-speciﬁed line, (g) ﬁnal result after ﬁlling in the remaining
unknown regions. (h
optimization. As shown in Fig.8, we want to ﬁll the missing geometry on the
Venus. The user speciﬁes the important missing texture structure by extending
a few curves from the known to the unknown regions on the base surface, the
signed gray-scale texture along the speciﬁed curves in the unknown region is
ﬁrst completed with structure guided synthesis and the other regions are then
completed by texture optimization.
6.3 Geometry Completion with Detail Cloning
Our method also provides a seamless geometry cloning tool for surface based
detail cloning. The missing region of one model can be completed with the geom-
etry details of other models as the geometry texture. The user speciﬁes a source
regionSin an arbitrary surface and the missing regionDon target surface.
By smoothing the source regionS, we get its signed gray-scale texture. Using
the constrained optimization texture completion technique described in above
sections, the signed gray-scale texture is adopted as the input texture sample to
complete the texture of the smooth patch ﬁlling holeD. After converting the
gray-scale texture back to the geometric detail, we obtain the cloning result as
shown in Fig.9. The defective region of model Lady is completed by transferring
the geometric details of the bob region on model Venus.

34 C. Xiao et al.
(a) (b (c (d
Fig. 9.Geometry completion based on geometry transfer. (a lladymodel,(b),
uncompleted lady model, (c) the sample model Venus, (d) the completion result.
7 Implementation and Results
The proposed algorithm was implemented on a Microsoft Windows XP PC with a
Celeron 2.00GHz CPU and 1.00GB RAM. The computational complexity of our
approach is dominated by the process of texture synthesis which is based on the
nearest neighbor search. In the example shown in Fig.3, there are 240,000 points
in the Stanford Bunny, the execution time per iteration takes 8-10 seconds, and
the total execution time is 10-12 minutes for about 50 iterations. In the context-
based geometry completion stage, the execution time is much less as the patched
region is much smaller. In ﬁg 6, there are 22,000 points in the patched regions,
it takes less than one minute to accomplish the texture completion operation. In
our experiments, there are usually about 300 points in each overlapped cluster,
the scale of the synthesized texture can be controlled by adjusting size of the
regular gridG.
8 Conclusion and Future Work
We have presented a novel approach for appearance and surface content com-
pletion for the acquired 3D data set based on global optimization. We transform
the task of surface content completion into that of surface texture completion.
The major beneﬁt is that it is ﬂexible and eﬃcient to implement. Our system
can be extended to the mesh models easily. Further research will be focused
on performing a user controllable non-uniform clustering so that the scale of
synthesized texture can vary progressively through the point cloud.
Acknowledgement
The work is supported by the National Grand Fundamental Research 973 Pro-
gram of China (No.2002CB312101)and the National Natural Science Foundation
of China under Grant Nos.60503056.

Random documents with unrelated
content Scribd suggests to you:

union of one drop of water with another, forming one globule. As
unconsciously and as positively are men constrained to band
together into societies as are particles forced to unite and form
crystals. And herein is a law as palpable and as fixed as any law in
nature; a law, which if unfulfilled, would result in the extermination
of the race. But the law of human attraction is not perfect, does not
fulfill its purpose apart from the law of human repulsion, for as we
have seen, until war and despotism and superstition and other dire
evils come, there is no progress. Solitude is insupportable, even
beasts will not live alone; and men are more dependent on each
other than beasts. Solitude carries with it a sense of inferiority and
insufficiency; the faculties are stinted, lacking completeness,
whereas volume is added to every individual faculty by union.
COÖPERATION AND THE DIVISION OF LABOR.
But association simply, is not enough; nothing materially great can
be accomplished without union and coöperation. It is only when
aggregations of families intermingle with other aggregations, each
contributing its quota of original knowledge to the other; when the
individual gives up some portion of his individual will and property
for the better protection of other rights and property; when he
entrusts society with the vindication of his rights; when he depends
upon the banded arm of the nation, and not alone upon his own arm
for redress of grievances, that progress is truly made. And with
union and coöperation comes the division of labor by which means
each, in some special department, is enabled to excel. By fixing the
mind wholly upon one thing, by constant repetition and practice, the
father hands down his art to the son, who likewise, improves it for
his descendants. It is only by doing a new thing, or by doing an old
thing better than it has ever been done before, that progress is
made. Under the régime of universal mediocrity the nation does not
advance; it is to the great men,—great in things great or small, that
progress is due; it is to the few who think, to the few who dare to

face the infinite universe of things and step, if need be, outside an
old-time boundary, that the world owes most.
Originally implanted is the germ of intelligence, at the first but little
more than brute instinct. This germ in unfolding undergoes a double
process; it throws off its own intuitions and receives in return those
of another. By an interchange of ideas, the experiences of one are
made known for the benefit of another, the inventions of one are
added to the inventions of another; without intercommunication of
ideas the intellect must lie dormant. Thus it is with individuals, and
with societies it is the same. Acquisitions are eminently reciprocal. In
society, wealth, art, literature, polity, and religion act and react on
each other; in science a fusion of antagonistic hypotheses is sure to
result in important developments. Before much progress can be
made, there must be established a commerce between nations for
the interchange of aggregated human experiences, so that the arts
and industries acquired by each may become the property of all the
rest, and thus knowledge becomes scattered by exchange, in place
of each having to work out every problem for himself. Thus viewed,
civilization is a partnership entered into for mutual improvement; a
joint stock operation, in which the product of every brain contributes
to a general fund for the benefit of all. No one can add to his own
store of knowledge without adding to the general store; every
invention, and discovery, however insignificant, is a contribution to
civilization.
In savagism, union and coöperation are imperfectly displayed. The
warriors of one tribe unite against the warriors of another; a band
will coöperate in pursuing a herd of buffalo; even one nation will
sometimes unite with another nation against a third, but such
combinations are temporary, and no sooner is the particular object
accomplished than the confederation disbands, and every man is
again his own master. The moment two or more persons unite for
the accomplishment of some purpose which shall tend permanently
to meliorate the condition of themselves and others, that moment
progress begins. The wild beasts of the forest, acting in unison, were

physically able to rise up and extirpate primitive man, but could
beasts in reality confederate and do this, such confederation of wild
beasts could become civilized.
THE SAVAGE HATES CIVILIZATION.
But why does primitive man desire to abandon his original state and
set out upon an arduous never-ending journey? Why does he wish
to change his mild paternal government, to relinquish his title to
lands as broad as his arm can defend, with all therein contained, the
common property of his people? Why does he wish to give up his
wild freedom, his native independence, and place upon his limbs the
fetters of a social and political despotism? He does not. The savage
hates civilization as he hates his deadliest foe; its choicest benefits
he hates more than the direst ills of his own unfettered life. He is
driven to it; driven to it by extraneous influences, without his
knowledge and against his will; he is driven to it by this Soul of
Progress. It is here that this progressional phenomenon again
appears outside of man and in direct opposition to the will of man; it
is here that the principle of evil again comes in and stirs men up to
the accomplishment of a higher destiny. By it Adam, the first of
recorded savages, was driven from Eden, where otherwise he would
have remained forever, and remained uncivilized. By it our ancestors
were impelled to abandon their simple state, and organize more
heterogeneous complex forms of social life. And it is a problem for
each nation to work out for itself. Millions of money are expended for
merely proselyting purposes, when if the first principles of civilization
were well understood, a more liberal manner of teaching would
prevail.
Every civilization has its peculiarities, its idiosyncrasies. Two
individuals attempting the same thing differ in the performance; so
civilization evolving under incidental and extraneous causes takes an
individuality in every instance. This is why civilizations will not
coalesce; this is why the Spaniards could make the Aztecs accept
their civilization only at the point of the sword. Development

engendered by one set of phenomena will not suit the developments
of other circumstances. The government, religion, and customs of
one people will not fit another people any more than the coat of one
person will suit the form of another. Thought runs in different
channels; the happiness of one is not the happiness of another;
development springs from inherent necessity, and one species
cannot be engrafted on another.
Let us now examine the phenomena of government and religion in
their application to the evolution of societies, and we shall better
understand how the wheels of progress are first set in motion,—and
by religion I do not mean creed or credulity, but that natural cultus
inherent in humanity, which is a very different thing. Government is
early felt to be a need of society; the enforcement of laws which
shall bring order out of social chaos; laws which shall restrain the
vicious, protect the innocent, and punish the guilty; which shall act
as a shield to inherent budding morality. But before government,
there must arise some influence which will band men together. An
early evil to which civilization is indebted is war; the propensity of
man—unhappily not yet entirely overcome—for killing his fellow-
man.
GOVERNMENT AND RELIGION.
The human race has not yet attained that state of homogeneous
felicity which we sometimes imagine; upon the surface, we yet bear
many of the relics of barbarism; under cover of manners, we hide
still more. War is a barbarism which civilization only intensifies, as
indeed civilization intensifies every barbarism which it does not
eradicate or cover up. The right of every individual to act as his own
avenger; trial by combat; justice dependent upon the passion or
caprice of the judge or ruler and not upon fixed law; hereditary
feuds and migratory skirmishes; these and the like are deemed
barbarous, while every nation of the civilized world maintains a

standing army, applies all the arts and inventions of civilization to the
science of killing, and upon sufficient provocation, as a disputed
boundary or a fancied insult, no greater nor more important than
that which moved our savage ancestors to like conduct, falls to, and
after a respectable civilized butchery of fifty or a hundred thousand
men, ceases fighting, and returns, perhaps, to right and reason as a
basis for the settlement of the difficulty. War, like other evils which
have proved instruments of good, should by this time have had its
day, should have served its purpose. Standing armies, whose
formation was one of the first and most important steps in
association and partition of labor, are but the manifestation of a
lingering necessity for the use of brute force in place of moral force
in the settlement of national disputes. Surely, rational beings who
retain the most irrational practices concerning the simplest principles
of social life cannot boast of a very high order of what we are
pleased to call civilization. Morality, commerce, literature, and
industry, all that tends toward elevation of intellect, is directly
opposed to the warlike spirit. As intellectual activity increases, the
taste for war decreases, for an appeal to war in the settlement of
difficulties is an appeal from the intellectual to the physical, from
reason to brute force.
Despotism is an evil, but despotism is as essential to progress as any
good. In some form despotism is an inseparable adjunct of war. An
individual or an idea may be the despot, but without cohesion,
without a strong central power, real or imaginary, there can be no
unity, and without unity no protracted warfare. In the first stages of
government despotism is as essential as in the last it is noxious. It
holds society together when nothing else would hold it, and at a
time when its very existence depends upon its being so held. And
not until a moral inherent strength arises sufficient to burst the
fetters of despotism, is a people fit for a better or milder form of
government; for not until this inherent power is manifest is there
sufficient cohesive force in society to hold it together without being
hooped by some such band as despotism. Besides thus cementing
society, war generates many virtues, such as courage, discipline,

obedience, chivalrous bearing, noble thought; and the virtues of war,
as well as its vices, help to mould national character.
Slavery to the present day has its defenders, and from the first it has
been a preventive of a worse evil,—slaughter. Savages make slaves
of their prisoners of war, and if they do not preserve them for slaves
they kill them. The origin of the word, servus, from servare, to
preserve, denotes humane thought rather than cruelty. Discipline is
always necessary to development, and slavery is another form of
savage discipline. Then, by systems of slavery, great works were
accomplished, which, in the absence of arts and inventions, would
not have been possible without slavery. And again, in early societies
where leisure is so necessary to mental cultivation and so difficult to
obtain, slavery, by promoting leisure, aids elevation and refinement.
Slaves constitute a distinct class, devoted wholly to labor, thereby
enabling another class to live without labor, or to labor with the
intellect rather than with the hands.
Primordially, society was an aggregation of nomadic families, every
head of a family having equal rights, and every individual such
power and influence as he could acquire and maintain. In all the
ordinary avocations of savage life this was sufficient; there was room
for all, and the widest liberty was possessed by each. And in this
happy state does mankind ever remain until forced out of it. In unity
and coöperation alone can great things be accomplished; but men
will not unite until forced to it. Now in times of war—and with
savages war is the rule and not the exception—some closer union is
necessary to avoid extinction; for other things being equal, the
people who are most firmly united and most strongly ruled are sure
to prevail in war. The idea of unity in order to be effectual must be
embodied in a unit; some one must be made chief, and the others
must obey, as in a band of wild beasts that follow the one most
conspicuous for its prowess and cunning. But the military principle
alone would never lay the foundation of a strong government, for
with every cessation from hostilities there would be a corresponding
relaxation of government.

GOVERNMENT FORCED UPON MAN.
Another necessity for government here arises, but which likewise is
not the cause of government, for government springs from force and
not from utility. These men do not want government, they do not
want culture; how then is an arm to be found sufficiently strong to
bridle their wild passions? In reason they are children, in passion
men; to restrain the strong passions of strong non-reasoning men
requires a power; whence is this power to come? It is in the earlier
stage of government that despotism assumes its most intense forms.
The more passionate, and lawless, and cruel the people, the more
completely do they submit to a passionate, lawless, and cruel prince;
the more ungovernable their nature, the more slavish are they in
their submission to government; the stronger the element to be
governed, the stronger must be the government.
The primitive man, whoever or whatever that may be, lives in
harmony with nature; that is, he lives as other animals live, drawing
his supplies immediately from the general storehouse of nature. His
food he plucks from a sheltering tree, or draws from a sparkling
stream, or captures from a prolific forest. The remnants of his
capture, unfit for food, supply his other wants; with the skin he
clothes himself, and with the bones makes implements and points
his weapons. In this there are no antagonisms, no opposing
principles of good and evil; animals are killed not with a view of
extermination, but through necessity, as animals kill animals in order
to supply actual wants. But no sooner does the leaven of progress
begin to work than war is declared between man and nature. To
make room for denser populations and increasing comforts, forests
must be hewn down, their primeval inhabitants extirpated or
domesticated, and the soil laid under more direct contribution. Union
and coöperation spring up for purposes of protection and
aggression, for the accomplishment of purposes beyond the capacity
of the individual. Gradually manufactures and commerce increase;
the products of one body of laborers are exchanged for the products
of another, and thus the aggregate comforts produced are doubled
to each. Absolute power is taken from the hands of the many and

placed in the hands of one, who becomes the representative power
of all. Men are no longer dependent upon the chase for a daily
supply of food; even agriculture no longer is a necessity which each
must follow for himself, for the intellectual products of one person or
people may be exchanged for the agricultural products of another.
With these changes of occupation new institutions spring up, new
ideas originate, and new habits are formed. Human life ceases to be
a purely material existence; another element finds exercise, the
other part of man is permitted to grow. The energies of society now
assume a different shape; hitherto the daily struggle was for daily
necessities, now the accumulation of wealth constitutes the chief
incentive to labor. Wealth becomes a power and absorbs all other
powers. The possessor of unlimited wealth commands the products
of every other man's labor.
But in time, and to a certain extent, a class arises already possessed
of wealth sufficient to satisfy even the demands of avarice, and
something still better, some greater good is yet sought for. Money-
getting gives way before intellectual cravings. The self-denials and
labor necessary to the acquisition of wealth are abandoned for the
enjoyment of wealth already acquired and the acquisition of a yet
higher good. Sensual pleasure yields in a measure to intellectual
pleasure, the acquisition of money to the acquisition of learning.
Where brute intelligence is the order of the day, man requires no
more governing than brutes, but when lands are divided, and the
soil cultivated, when wealth begins to accumulate and commerce
and industry to flourish, then protection and lawful punishment
become necessary. Like the wild horse, leave him free, and he will
take care of himself; but catch him and curb him, and the wilder and
stronger he is the stronger must be the curb until he is subdued and
trained, and then he is guided by a light rein. The kind of
government makes little difference so that it be strong enough.
THE SUPERNATURAL IN CIVILIZATION.

Granted that it is absolutely essential to the first step toward culture
that society should be strongly governed, how is the first
government to be accomplished; how is one member of a
passionate, unbridled heterogeneous community to obtain dominion
absolute over all the others? Here comes in another evil to the
assistance of the former evils, all for future good,—Superstition.
Never could physical force alone compress and hold the necessary
power with which to burst the shell of savagism. The government is
but a reflex of the governed. Not until one man is physically or
intellectually stronger than ten thousand, will an independent people
submit to a tyrannical government, or a humane people submit to a
cruel government, or a people accustomed to free discussion to an
intolerant priesthood.
At the outset, if man is to be governed at all, there must be no
division of governmental force. The cause for fear arising from both
the physical and the supernatural must be united in one individual.
In the absence of the moral sentiment the fear of legal and that of
spiritual punishments are identical, for the spiritual is feared only as
it works temporal or corporal evil. Freedom of thought at this stage
is incompatible with progress, for thought without experience is
dangerous, tending towards anarchy. Before men can govern
themselves they must be subjected to the sternest discipline of
government, and whether this government be just or humane or
pleasant is of small consequence so that it be only strong enough.
As with polity so with morality and religion; conjointly with
despotism there must be an arbitrary central church government, or
moral anarchy is the inevitable consequence. At the outset it is not
for man to rule but to obey; it is not for savages, who are children in
intellect to think and reason, but to believe.
And thus we see how wonderfully man is provided with the
essentials of growth. This tender germ of progress is preserved in
hard shells and prickly coverings, which, when they have served
their purpose are thrown aside as not only useless but detrimental to
further development. We know not what will come hereafter, but up

to the present time a state of bondage appears to be the normal
state of humanity; bondage, at first severe and irrational, then ever
loosening, and expanding into a broader freedom. As mankind
progresses, moral anarchy no more follows freedom of thought than
does political anarchy follow freedom of action. In Germany, in
England, in America, wherever secular power has in any measure cut
loose from ecclesiastical power and thrown religion back upon public
sentiment for support, a moral as well as an intellectual advance has
always followed. What the mild and persuasive teachings and lax
discipline of the present epoch would have been to the Christians of
the fourteenth century, the free and lax government of republican
America would have been to republican Rome. Therefore, let us
learn to look charitably upon the institutions of the past, and not
forget how much we owe to them; while we rejoice at our release
from the cruelty and ignorance of mediæval times, let us not forget
the debt which civilization owes to the rigorous teachings of both
Church and State.
MORALITY AND CREED.
Christianity, by its exalted un-utilitarian morality and philanthropy,
has greatly aided civilization. Indeed so marked has been the effect
in Europe, so great the contrast between Christianity and Islamism
and the polytheistic creeds in general, that Churchmen claim
civilization as the offspring of their religion. But religion and morality
must not be confounded with civilization. All these and many other
activities act and react on each other as proximate principles in the
social organism, but they do not, any or all of them, constitute the
life of the organism. Long before morality is religion, and long after
morality religion sends the pious penitent to his knees. Religious
culture is a great assistant to moral culture as intellectual training
promotes the industrial arts, but morality is no more religion than is
industry intellect. When Christianity, as in the early settlement of
Mexico and Central America, falls into the hands of unprincipled
adventurers or blind zealots who stand up in deadly antagonism to

liberty, then Christianity is a drag upon civilization; and therefore we
may conclude that in so far as Christianity grafts on its code of pure
morality the principle of intellectual freedom, in so far is civilization
promoted by Christianity, but when Christianity engenders
persecution, civilization is retarded thereby.
Then Protestantism sets up a claim to the authorship of civilization,
points to Spain and then to England, compares Italy and
Switzerland, Catholic America and Puritan America, declares that the
intellect can never attain superiority while under the dominion of the
Church of Rome; in other words, that civilization is Protestantism. It
is true that protestation against irrational dogmas, or any other
action that tends toward the emancipation of the intellect, is a great
step in advance; but religious belief has nothing whatever to do with
intellectual culture. Religion from its very nature is beyond the limits
of reason; it is emotional rather than intellectual, an instinct and not
an acquisition. Between reason and religion lies a domain of
common ground upon which both may meet and join hands, but
beyond the boundaries of which neither may pass. The moment the
intellect attempts to penetrate the domain of the Supernatural all
intellectuality vanishes, and emotion and imagination fill its place.
There can be no real conflict between the two, for neither, by any
possibility, can pass this neutral ground. Before the mind can receive
Christianity, or Mahometanism, or any other creed, it must be ready
to accept dogmas in the analysis of which human reason is
powerless. Among the most brilliant intellects are found Protestants,
Romanists, Unitarians, Deists, and Atheists; judging from the
experiences of mankind in ages past, creeds and formulas,
orthodoxy and heterodoxy, have no inherent power to advance or
retard the intellect. Some claim, indeed, that strong doctrinal bias
stifles thought, fosters superstition, and fetters the intellect; still
religious thought, in some form, is inseparable from the human
mind, and it would be very difficult to prove that belief is more
debasing than non-belief.

DEVELOPMENT OF THE RELIGIOUS IDEAL.
Religion at first is a gross fetichism, which endows every wonder
with a concrete personality. Within every appearance is a several
personal cause, and to embody this personal cause in some material
form is the first effort of the savage mind. Hence, images are made
in representation of these imaginary supernatural powers. Man, of
necessity, must clothe these supernatural powers in the elements of
some lower form. The imagination cannot grasp an object or an idea
beyond the realms of human experience. Unheard-of combinations
of character may be made, but the constituent parts must, at some
time and in some form, have had an existence in order to be
conceivable. It is impossible for the human mind to array in forms of
thought anything wholly and absolutely new. This state is the
farthest remove possible from a recognition of those universal laws
of causation toward which every department of knowledge is now so
rapidly tending. Gods are made in the likeness of man and beast,
endowed with earthly passions, and a sensual polytheism, in which
blind fate is a prominent element, becomes the religious ideal.
Religious conceptions are essentially material; all punishments and
rewards are such as effect man as a material being; morality, the
innate sense of right and wrong, lies stifled, almost dormant.
Thrown wholly upon himself, without experience to guide him, the
savage must, of necessity, invest nature with his own qualities, for
his mind can grasp none other. But when experience dispels the
nearer illusions, objects more remote are made gods; in the sun and
stars he sees his controlling destinies; the number of his gods is
lessened until at last all merge into one God, the author of all law,
the great and only ruler of the universe. In every mythology we see
this impersonation of natural phenomena; frost and fire, earth and
air and water, in their displays of mysterious powers, are at once
deified and humanized. These embodiments of physical force are
then naturally formed into families, and their supposed descendants
worshiped as children of the gods. Thus, in the childhood of society,
when incipient thought takes up its lodgment in old men's brains,

shadows of departed heroes mingle with shadows of mysterious
nature, and admiration turns to adoration.
Next arises the desire to propitiate these unseen powers, to
accomplish which some means of communication must be opened
up between man and his deities. Now, as man in his gods
reproduces himself, as all his conceptions of supernatural power
must, of necessity, be formed on the skeleton of human power,
naturally it follows that the strongest and most cunning of the tribe,
he upon whom leadership most naturally falls, comes to be regarded
as specially favored of the gods. Powers supernatural are joined to
powers temporal, and embodied in the chieftain of the nation. A
grateful posterity reveres and propitiates departed ancestors. The
earlier rulers are made gods, and their descendants lesser divinities;
the founder of a dynasty, perhaps, the supreme god, his progeny
subordinate deities. The priesthood and kingship thus become
united; religion and civil government join forces to press mankind
together, and the loose sands of the new strata cohere into the firm
rock, that shall one day bear alone the wash of time and tide.
Hence arise divine kingship, and the divine right of kings, and with
the desire to win the favor of this divine king, arise the courtesies of
society, the first step toward polish of manners. Titles of respect and
worship are given him, some of which are subsequently applied to
the Deity, while others drop down into the common-place
compliments of every-day life.
Here then, we have as one of the first essentials of progress the
union of Church and State, of superstition and despotism, a union
still necessarily kept up in some of the more backward civilizations.
Excessive loyalty and blind faith ever march hand in hand. The very
basis of association is credulity, blind loyalty to political powers and
blind faith in sacerdotal terrors. In all mythologies at some stage
temporal and spiritual government are united, the supernatural
power being incarnated in the temporal chief; political despotism
and an awful sanguinary religion,—a government and a belief, to
disobey which was never so much as thought possible.

See how every one of these primary essentials of civilization
becomes, as man advances, a drag upon his progress; see how he
now struggles to free himself from what, at the outset, he was led
by ways he knew not to endure so patiently. Government, in early
stages always strong and despotic, whether monarchical,
oligarchical, or republican, holding mankind under the dominion of
caste, placing restrictions upon commerce and manufactures,
regulating social customs, food, dress,—how men have fought to
break loose these bonds! Religion, not that natural cultus instinctive
in humanity, the bond of union as well under its most disgusting
form of fetichism, as under its latest, loveliest form of Christianity;
but those forms and dogmas of sect and creed which stifle thought
and fetter intellect,—how men have lived lives of sacrifice and self-
denial as well as died for the right to free themselves from
unwelcome belief!
RELATION OF GOVERNMENT TO CIVILIZATION.
In primeval ages, government and religion lay lightly on the human
race; ethnology, as well as history, discloses the patriarchal as the
earliest form of government, and a rude materialism as the earliest
religious ideal; these two simple elements, under the form of
monsters, became huge abortions, begotten of ignorance, that held
the intellect in abject slavery for thousands of years, and from these
we, of this generation, more than any other, are granted
emancipation. Even wealth, kind giver of grateful leisure, in the
guise of avarice becomes a hideous thing, which he who would
attain the higher intellectual life, must learn to despise.
Government, as we have seen, is not an essential element of
collective humanity. Civilization must first be awakened, must even
have passed the primary stages before government appears.
Despotism, feudalism, divine kingship, slavery, war, superstition,
each marks certain stages of development, and as civilization

advances all tend to disappear; and, as in the early history of
nations the state antedates the government, so the time may come
in the progress of mankind when government will be no longer
necessary. Government always grows out of necessity; the intensity
of government inevitably following necessity. The form of
government is a natural selection; its several phases always the
survival of the fittest. When the federalist says to the monarchist, or
the monarchist to the federalist: My government is better than
yours, it is as if the Eskimo said to the Kaffir: My coat, my house, my
food, is better than yours.
The government is made for the man, and not the man for the
government. Government is as the prop for the growing plant; at
first the young shoot stands alone, then in its rapid advancement for
a time it requires support, after which it is able again to stand alone.
What we term the evils of government are rather its necessities, and
are, indeed, no evils at all. The heavy bit which controls the mouth
of an untamed horse is to that horse an evil, yet to the driver a
necessity which may be laid aside as the temper of the animal is
subdued. So despotism, feudalism, slavery, are evils to those under
their dominion, yet are they as necessary for the prevention of
anarchy, for the restraint of unbridled passions, as the powerful bit
to the horse, and will as surely be laid aside when no longer
required. Shallow-minded politicians talk of kingcraft, arbitrary rule,
tyrants, the down-trodden masses, the withholding of just rights; as
though the government was some independent, adverse element,
wholly foreign to the character of the people; as though one man
was stronger than ten thousand; as though, if these phases of
society were not the fittest, they would be tolerated for a moment.
The days of rigorous rule were ever the best days of France and
Spain, and so it will be until the people become stronger than the
strength of rulers. Republicanism is as unfit for stupid and
unintellectual populations, as despotism would be for the advanced
ideas and liberal institutions of Anglo-Saxon America. The subject of
a liberal rule sneeringly crying down to the subject of an absolute
rule his form of government, is like the ass crying to the tiger: Leave

blood and meat; feed on grass and thistles, the only diet fit for
civilized beasts! Our federal government is the very best for our
people, when it is not so it will speedily change; it fits the temper of
American intelligence, but before it can be planted in Japan or China
the traditions and temper of the Asiatics must change.
We of to-day are undergoing an important epoch in the history of
civilization. Feudalism, despotism, and fanaticism have had each its
day, have each accomplished its necessary purpose, and are fast
fading away. Ours is the age of democracy, of scientific investigation,
and freedom of religious thought; what these may accomplish for
the advancing intellect remains to be seen. Our ancestors loved to
dwell upon the past, now we all look toward the future.
LATTER-DAY PROGRESSION.
The sea of ice, over which our forefathers glided so serenely in their
trustful reliance, is breaking up. One after another traditions
evaporate; in their application to proximate events they fail us,
history ceases to repeat itself as in times past. Old things are
passing away, all things are becoming new; new philosophies, new
religions, new sciences; the industrial spirit springs up and overturns
time-honored customs; theories of government must be
reconstructed. Thus, says experience, republicanism, as a form of
government, can exist only in small states; but steam and electricity
step in and annihilate time and space. The Roman republic, from a
lack of cohesive energy, from failure of central vital power sufficient
to send the blood of the nation from the heart to the extremities,
died a natural death. The American republic, covering nearly twice
the territory of republican Rome in her palmiest days, is endowed
with a different species of organism; in its physiological system is
found a new series of veins and arteries, the railway, the telegraph,
and the daily press,—through which pulsates the life's blood of the
nation, millions inhaling and exhaling intelligence as one man. By
means of these inventions all the world, once every day, are brought
together. By telegraphic wires and railroad iron men are now bound

as in times past they were bound by war, despotism, and
superstition. The remotest corners of the largest republics of to-day,
are brought into closer communication than were the adjoining
states of the smallest confederations of antiquity. A united Germany,
from its past history held to be an impossibility, is, with the present
facilities of communication, an accomplished fact. England could as
easily have possessed colonies in the moon, as have held her
present possessions, three hundred years ago. Practically, San
Francisco is nearer Washington than was Philadelphia when the
foundations of the Capitol were laid. What is to prevent republics
from growing, so long as intelligence keeps pace with extension?
The general of an army may now sit before his maps, and
manœuvre half a score of armies a hundred or a thousand miles
apart, know hourly the situation of every division, the success of
every battle, order an advance or a retreat, lay plots and make
combinations, with more exactness than was once possible in the
conduct of an ordinary campaign.
MORALS, MANNERS, AND FASHION.
A few words about morals, manners, and fashion, will further
illustrate how man is played upon by his environment, which here
takes the shape of habit. In their bearing on civilization, these
phenomena all come under the same category; and this, without
regard to the rival theories of intuition and utility in morals.
Experience teaches, blindly at first yet daily with clearer vision, that
right conduct is beneficial, and wrong conduct detrimental; that the
consequences of sin invariably rest on the evil-doer; that for an
unjust act, though the knowledge of it be forever locked in the
bosom of the offender, punishment is sure to follow; yet there are
those who question the existence of innate moral perceptions, and
call it all custom and training. And if we look alone to primitive
people for innate ideas of morality and justice I fear we shall meet
with disappointment. Some we find who value female chastity only

before marriage, others only after marriage,—that is, after the
woman and her chastity both alike become the tangible property of
somebody. Some kindly kill their aged parents, others their female
infants; the successful Apache horse-thief is the darling of his
mother, and the hero of the tribe; often these American Arabs will
remain from home half-starved for weeks, rather than suffer the
ignominy of returning empty-handed. Good, in the mind of the
savage, is when he steals wives; bad, is when his own wives are
stolen. Where it is that inherent morality in savages first makes its
appearance, and in what manner, it is often difficult to say; the most
hideous vices are everywhere practiced with unblushing effrontery.
Take the phenomena of Shame. Go back to the childhood of our
race, or even to our own childhood, and it will be hard to discover
any inherent quality which make men ashamed of one thing more
than another. Nor can the wisest of us give any good and sufficient
reason why we should be ashamed of our body any more than of
our face. The whole man was fashioned by one Creator, and all parts
equally are perfect and alike honorable. We cover our person with
drapery, and think thereby to hide our faults from ourselves and
others, as the ostrich hides its head under a leaf, and fancies its
body concealed from the hunter. What is this quality of shame if it
be not habit? A female savage will stand unblushingly before you
naked, but strip her of her ornaments and she will manifest the
same appearance of shame, though not perhaps so great in degree,
that a European woman will manifest if stripped of her clothes. It is
well known how civilized and semi-civilized nations regard this
quality of propriety. Custom, conventional usage, dress and behavior,
are influences as subtle and as strong as any that govern us,
weaving their net-work round man more and more as he throws off
allegiance to other powers; and we know but little more of their
origin and nature than we do of the origin and nature of time and
space, of life and death, of origin and end.
Every age and every society has its own standard of morality, holds
up some certain conduct or quality as a model, saying to all, Do this,

and receive the much-coveted praise of your fellows. Often what one
people deem virtue is to another vice; what to one age is religion is
to another superstition; but underlying all this are living fires, kindled
by Omnipotence, and destined to burn throughout all time. In the
Spartan and Roman republics the moral ideal was patriotism; among
mediæval Churchmen it took the form of asceticism; after the
elevation of woman the central idea was female chastity.
In this national morality, which is the cohesive force of the body
social, we find the fundamental principle of the progressional
impulse, and herein is the most hopeful feature of humanity;
mankind must progress, and progress in the right direction. There is
no help for it until God changes the universal order of things; man
must become better in spite of himself; it is the good in us that
grows and ultimately prevails.
As a race we are yet in our nonage; fearful of the freedom given us
by progress we cling tenaciously to our leading-strings; hugging our
mother, Custom, we refuse to be left alone. Liberty and high
attainments must be meted out to us as we are able to receive
them, for social retchings and vomitings inevitably follow over-
feedings. Hence it is, that we find ourselves escaped from primeval
and mediæval tyrannies only to fall under greater ones; society is
none the less inexorable in her despotisms because of the sophistry
which gives her victims fancied freedom. For do we not now set up
forms and fashions, the works of our own hands, and bow down to
them as reverently as ever our heathen ancestors did to their gods
of wood and stone? Who made us? is not the first question of our
catechism, but What will people say?
ORIGIN AND SIGNIFICANCE OF DRESS.
Of all tyrannies, the tyranny of fashion is the most implacable; of all
slaveries the slavery to fashion is the most abject; of all fears the
fear of our fellows is the most overwhelming; of all the influences

that surround and govern man the forms and customs which he
encounters in society are the most domineering. It is the old story,
only another turn of the wheel that grinds and sharpens and
polishes humanity,—at the first a benefit, now a drag. Forms and
fashions are essential; we cannot live without them. If we have
worship, government, commerce, or clothes, we must have forms;
or if we have them not we still must act and do after some fashion;
costume, which is but another word for custom, we must have, but
is it necessary to make the form the chief concern of our lives while
we pay so little heed to the substance? and may we not hope while
rejoicing over our past emancipations, that we shall some day be
free from our present despotisms?
Dress has ever exercised a powerful influence on morals and on
progress; but this vesture-phenomenon is a thing but imperfectly
understood. Clothes serve as a covering to the body of which we are
ashamed, and protect it against the weather, and these, we infer, are
the reasons of our being clothed. But the fact is, aboriginally, except
in extreme cases, dress is not essential to the comfort of man until it
becomes a habit, and as for shame, until told of his nakedness, the
primitive man has none. The origin of dress lies behind all this; it is
found in one of the most deep-rooted elements of our nature,
namely, in our love of approbation. Before dress is decoration. The
successful warrior, proud of his achievement, besmears his face and
body with the blood of the slain, and straightway imitators, who also
would be thought strong and brave, daub themselves in like manner;
and so painting and tattooing become fashionable, and pigments
supply the place of blood. The naked, houseless Californian would
undergo every hardship, travel a hundred miles, and fight a round
with every opposing band he met, in order to obtain cinnabar from
the New Almaden quicksilver mine. So when the hunter kills a wild
beast, and with the tail or skin decorates his body as a trophy of his
prowess, others follow his example, and soon it is a shame to that
savage who has neither paint, nor belt, nor necklace of bears' claws.
And so follow head-flattenings, and nose-piercings, and lip-cuttings,
and, later, chignons, and breast-paddings, and bustles. Some say

that jealousy prompted the first Benedicks to hide their wives'
charms from their rivals, and so originated female dress, which, from
its being so common to all aborigines, is usually regarded as the
result of innate modesty. But whatever gave us dress, dress has
given much to human progress. Beneath dress arose modesty and
refinement, like the courtesies that chivalry threw over feudalism,
covering the coarse brutality of the barons, and paving the way to
real politeness.
ETIQUETTE, MORALITY, LAWS.
From the artificial grimaces of fashion have sprung many of the
natural courtesies of life; though here, too, we are sent back at once
to the beginning for the cause. From the ages of superstition and
despotism have descended the expressions of every-day politeness.
Thus we have sir, from sieur, sire, seigneur, signifying ruler, king,
lord, and aboriginally father. So madam, ma dame, my lady, formerly
applied only to women of rank. In place of throwing ourselves upon
the ground, as before a god or prince, we only partially prostrate
ourselves in bowing, and the hat which we touch to an acquaintance
we take off on entering a church in token of our humility. Again, the
captive in war is made a slave, and as such is required to do
obeisance to his master, which forms of servility are copied by the
people in addressing their superiors, and finally become the
established usage of ordinary intercourse. Our daily salutations are
but modified acts of worship, and our parting word a benediction;
and from blood, tomahawks, and senseless superstitions we turn
and find all the world of humanity, with its still strong passions and
subtle cravings, held in restraint by a force of which its victims are
almost wholly unconscious,—and this force is Fashion. In tribunals of
justice, in court and camp etiquette, everywhere these relics of
barbarism remain with us. Even we of this latter-day American
republicanism, elevate one of our fellows to the chieftainship of a
federation or state, and call him Excellency; we set a man upon the
bench and plead our cause before him; we send a loafer to a

legislature, and straightway call him Honorable,—such divinity doth
hedge all semblance of power.
Self-denial and abstinence lie at the bottom of etiquette and good
manners. If you would be moral, says Kant, you must "act always so
that the immediate motive of thy will may become a universal rule
for all intelligent beings," and Goethe teaches that "there is no
outward sign of courtesy that does not rest on a deep, moral
foundation."
Fine manners, though but the shell of the individual, are, to society,
the best actions of the best men crystallized into a mode; not only
the best thing, but the best way of doing the best thing. Good
society is, or ought to be, the society of the good; but fashion is
more than good society, or good actions; it is more than wealth, or
beauty, or genius, and so arbitrary in its sway that, not unfrequently,
the form absorbs the substance, and a breach of decorum becomes
a deadly sin.
Thus we see in every phase of development the result of a social
evolution; we see men coming and going, receiving their leaven
from the society into which by their destiny they are projected, only
to fling it back into the general fund interpenetrated with their own
quota of force. Meanwhile, this aggregation of human experiences,
this compounding of age with age, one generation heaping up
knowledge upon another; this begetting of knowledge by
knowledge, the seed so infinitesimal, the tree now so rapidly sending
forth its branches, whither does it tend? Running the eye along the
line of progress, from the beginning to the end, the measure of our
knowledge seems nearly full; resolving the matter, experience
assures us that, as compared with those who shall come after us,
we are the veriest barbarians. The end is not yet; not until infinity is
spanned and eternity brought to an end, will mankind cease to
improve.

Out of this conglomeration of interminable relationships concordant
and antagonistic laws are ever evolving themselves. Like all other
progressional phenomena, they wait not upon man; they are self-
creative, and force themselves upon the mind age after age, slowly
but surely, as the intellect is able to receive them; laws without law,
laws unto themselves, gradually appearing as from behind the mists
of eternity. At first, man and his universe appear to be regulated by
arbitrary volitions, by a multitude of individual minds; each governs
absolutely his own actions; every phenomenon of nature is but the
expression of some single will. As these phenomena, one after
another, become stripped of their mystery, there stands revealed not
a god, but a law; seasons come and go, and never fail; sunshine
follows rain, not because a pacified deity smiles, but because the
rain-clouds have fallen and the sun cannot help shining. Proximate
events first are thus made godless, then the whole host of deities is
driven farther and farther back. Finally the actions of man himself
are found to be subject to laws. Left to his own will, he wills to do
like things under like conditions.
As to the nature of these laws, the subtle workings of which we see
manifest in every phase of society, I cannot even so much as speak.
An infinite ocean of phenomena awaits the inquirer; an ocean
bottomless, over whose surface spreads an eternity of progress, and
beneath whose glittering waves the keenest intellect can scarcely
hope to penetrate far. The universe of man and matter must be
anatomized; the functions of innumerable and complex organs
studied; the exercise and influence of every part on every other part
ascertained, and events apparently the most capricious traced to
natural causes; then, when we know all, when we know as God
knoweth, shall we understand what it is, this Soul of Progress.

CHAPTER II.
GENERAL VIEW OF THE CIVILIZED NATIONS.
ThÉ AmÉrácan Cáválázaíáon of íhÉ SáxíÉÉníh CÉníury—Iíë DáëaééÉarancÉ —ThÉ Paëí, a NÉw
ElÉmÉní —Dávádáng lánÉ bÉíwÉÉn SavagÉ and CáválázÉd TrábÉë—Boundë of AmÉrácan
Cáválázaíáon—Phyëácal FÉaíurÉë of íhÉ Couníry—Maya and Nahua BranchÉë of
Aborágánal CulíurÉ—ThÉ Nahua Cáválázaíáon—ThÉ AzíÉcë áíë RÉérÉëÉníaíávÉë—Lámáíë of
íhÉ AzíÉc EméárÉ—AncáÉní Háëíory of Anáhuac án OuílánÉ —ThÉ TolíÉc Era—ThÉ
CháchámÉc Era—ThÉ AzíÉc Era—ExíÉní of íhÉ AzíÉc LanguagÉ—CáválázÉd PÉoélÉë
ouíëádÉ of Anáhuac—CÉníral AmÉrácan Naíáonë—ThÉ Maya CulíurÉ—ThÉ PrámáíávÉ
Maya EméárÉ—Nahua InfluÉncÉ án íhÉ Souíh—Yucaían and íhÉ Mayaë—ThÉ Naíáonë of
Cháaéaë—ThÉ Quáché EméárÉ án GuaíÉmala—ThÉ Nahuaë án Nácaragua and Salvador—
Eíymology of NamÉë.
In the preceding volume I have had occasion several times to
remark that, in the delineation of the Wild Tribes of the Pacific
States, no attempt is made to follow them in their rapid decline, no
attempt to penetrate their past or prophesy a possible future, no
profitless lingering over those misfortunes that wrought among them
such swift destruction. To us the savage nations of America have
neither past nor future; only a brief present, from which indeed we
may judge somewhat of their past; for the rest, foreign avarice and
interference, European piety and greed, saltpetre, steel, small-pox,
and syphilis, tell a speedy tale. Swifter still must be the hand that

sketches the incipient civilization of the Mexican and Central
American table-lands. For although here we have more past, there is
still less present, and scarcely any future. Those nations raised the
highest by their wealth and culture, were the first to fall before the
invader, their superior attainments offering a more shining mark to a
rapacious foe; and falling, they were the soonest lost,—absorbed by
the conquering race, or disappearing in the surrounding darkness.
Although the savage nations were rapidly annihilated, traces of
savagism lingered, and yet linger; but the higher American culture, a
plant of more delicate growth and more sensitive nature, withered at
the first rude touch of foreign interference. Instead of being left to
its own intuitive unfoldings, or instead of being fostered by the new-
comers, who might have elevated by interfusion both their own
culture and that of the conquered race, the spirit of progress was
effectually stifled on both sides by fanatical attempts to substitute by
force foreign creeds and polities for those of indigenous origin and
growth. And now behold them both, the descendants of conquerors
and of conquered, the one scarcely less denaturalized than the other,
the curse inflicted by the invaders on a flourishing empire returning
and resting with crushing weight on their own head. Scarce four
centuries ago the empire of Charles the Fifth, and the empire of
Montezuma the Second, were brought by the force of progress most
suddenly and unexpectedly face to face; the one then the grandest
and strongest of the old world as was the other of the new. Since
which time the fierce fanaticism that overwhelmed the New World
empire, has pressed like an incubus upon the dominant race, and
held it fast while all the world around were making the most rapid
strides forward.
THE PAST, A NEW ELEMENT.
No indigenous civilization exists in America to-day, yet the effects of
a former culture are not altogether absent. The descendant of the
Aztec, Maya, and Quiché, is still of superior mind and haughtier spirit
than his roving brother who boasts of none but a savage ancestry.

Still, so complete has been the substitution of foreign civil and
ecclesiastical polities, and so far-reaching their influence on native
character and conduct; so intimate the association for three and
more centuries with the Spanish element; so closely guarded from
foreign gaze has been every manifestation of the few surviving
sparks of aboriginal modes of thought, that a study of the native
condition in modern times yields, by itself, few satisfactory results.
This study, however, as part of an investigation of their original or
normal condition, should by no means be neglected, since it may
furnish illustrative material of no little value.
Back of all this lies another element which lends to our subject yet
grander proportions. Scattered over the southern plateaux are heaps
of architectural remains and monumental piles. Furthermore, native
traditions, both orally transmitted and hieroglyphically recorded by
means of legible picture-writings, afford us a tolerably clear view of
the civilized nations during a period of several centuries preceding
the Spanish conquest, together with passing glances, through
momentary clearings in the mythologic clouds, at historical epochs
much more remote. Here we have as aids to this analysis,—aids
almost wholly wanting among the so-called savage tribes,
antiquities, tradition, history, carrying the student far back into the
mysterious New World past; and hence it is that from its
simultaneous revelation and eclipse, American civilization would
otherwise offer a more limited field for investigation than American
savagism, yet by the introduction of this new element the field is
widely extended.
Nor have we even yet reached the limits of our resources for the
investigation of this New World civilization. In these relics of
architecture and literature, of mythology and tradition, there are
clear indications of an older and higher type of culture than that
brought immediately to the knowledge of the invaders; of a type
that had temporarily deteriorated, perhaps through the influence of
long-continued and bloody conflicts, civil and foreign, by which the
more warlike rather than the more highly cultured nations had been

brought into prominence and power. But this anterior and superior
civilization, resting largely as it does on vague tradition, and
preserved to our knowledge in general allusions rather than in detail,
may, like the native condition since the conquest, be utilized to the
best advantage here as illustrative of the later and better-known, if
somewhat inferior civilization of the sixteenth century, described by
the conqueror, the missionary, and the Spanish historian.
Antique remains of native skill, which have been preserved for our
examination, may also be largely used in illustration of more modern
art, whose products have disappeared. These relics of the past are
also of the highest value as confirming the truth of the reports made
by Spanish writers, very many, or perhaps most, of whose
statements respecting the wonderful phenomena of the New World,
without this incontrovertible material proof, would find few believers
among the sceptical students of the present day. These remains of
antiquity, however, being fully described in another volume of this
work, may be referred to in very general terms for present purposes.
ORIGIN OF AMERICAN CIVILIZATION.
Of civilization in general, the nature of its phenomena, the causes
and processes by which it is evolved from savagism, I have spoken
sufficiently in the foregoing chapter. As for the many theories
respecting the American civilization in particular, its origin and
growth, it is not my purpose to discuss them in this volume. No
theory on these questions could be of any practical value in the
elucidation of the subject, save one that should stand out among the
rest so preëminently well-founded as to be generally accepted
among scientific men, and no one of all the multitude proposed has
acquired any such preëminence. A complete résumé of all the
theories on the subject, with the foundations which support them, is
given elsewhere in connection with the ancient traditionary history of
the aboriginal nations. It is well, however, to remark that our lack of
definite knowledge about the origin of this civilization is not
practically so important as might appear at first thought. True, we

know not for certain whether it is indigenous or exotic; and if the
former, whether to ascribe its cradle to the north or south, to one
locality or many; or if the latter, whether contact with the old world
was effected at one or many points, on one occasion or at divers
epochs, through the agency of migrating peoples or by the advent of
individual civilizers and teachers. Yet the tendency of modern
research is to prove the great antiquity of the American civilization
as well as of the American people; and if either was drawn from a
foreign source, it was at a time probably so remote as to antedate
any old-world culture now existing, and to prevent any light being
thrown on the offspring by a study of the parent stock; while if
indigenous, little hope is afforded of following rationally their
development through the political convulsions of the distant past
down to even a traditionally historic epoch.
I may then dispense with theories of origin and details of past
history as confusing rather than aiding my present purpose, and as
being fully treated elsewhere in this work. Neither am I required in
this treatment of the civilized races to make an accurate division
between them and their more savage neighbors, to determine the
exact standard by which savagism and civilization are to be
measured, or to vindicate the use of the word civilized as applied to
the American nations in preference to that of semi-civilized,
preferred by many writers. We have seen that civilization is at best
only a comparative term, applied to some of the ever-shifting phases
of human progress. In many of the Wild Tribes already described
some of its characteristics have been observed, and the opposite
elements of savagism will not be wanting among what I proceed to
describe as the Civilized Nations. There is not a savage people
between Anáhuac and Nicaragua that has not been influenced in its
institutions by intercourse, warlike, social, or commercial, with
neighbors of higher culture, and has not exerted in its turn a reflex
influence on the latter. The difficulty of drawing division-lines
between nations thus mutually acting on each other is further
increased in America by the fact that two or three nations constitute
the central figure of nearly all that has been observed or written by

the few that came in actual contact with the natives. This volume
will, therefore, deal rather with the native civilization than with the
nations that possessed it.
While, however, details on all the points mentioned, outside of actual
institutions found existing in the sixteenth century, would tend to
confusion rather than to clearness, besides leading in many cases to
endless repetition, yet a general view of the whole subject, of the
number, extent, location, and mutual relations of the nations
occupying the central portions of the continent at its discovery, as
well as of their relations to those of the more immediate past,
appears necessary to an intelligent perusal of the following pages. In
this general view I shall avoid all discussion of disputed questions,
reserving arguments and details for future volumes on antiquities
and aboriginal history.
HOME OF THE AMERICAN CULTURE.
That portion of what we call the Pacific States which was the home
of American civilization within historic or traditionally historic times,
extends along the continent from north-west to south-east, between
latitudes 22° and 11°. On the Atlantic side the territory stretches
from Tamaulipas to Honduras, on the Pacific from Colima to
Nicaragua. Not that these are definitely drawn boundaries, but
outside of these limits, disregarding the New Mexican Pueblo culture,
this civilization had left little for Europeans to observe, while within
them lived few tribes uninfluenced or unimproved by contact with it.
No portion of the globe, perhaps, embraces within equal latitudinal
limits so great a variety of climate, soil, and vegetation; a variety
whose important bearing on the native development can be
understood in some degree, and which would doubtless account
satisfactorily for most of the complications of progressional
phenomena observed within the territory, were the connection
between environment and progress fully within the grasp of our

knowledge. All the gradations from a torrid to a temperate clime are
here found in a region that lies wholly within the northern tropic,
altitudinal variations taking the place of and producing all the effects
elsewhere attributable to latitude alone. These variations result from
the topography of the country as determined by the conformation
given to the continent by the central cordillera. The Sierra Madre
enters this territory from the north in two principal ranges, one
stretching along the coast of the Pacific, while the other and more
lofty range trends nearer the Atlantic, the two again uniting before
reaching the isthmus of Tehuantepec. This eastern branch between
18° 40´ and 20° 30´ opens out into a table-land of some seventy-
five by two hundred miles area, with an altitude of from six to eight
thousand feet above the sea level. This broad plateau or series of
plateaux is known as the tierra fria, while the lower valleys, with a
band of the surrounding slopes, at an elevation of from three to five
thousand feet, including large portions of the western lands of
Michoacan, Guerrero, and Oajaca, between the two mountain
branches, constitute the tierra templada. From the surface of the
upper table-land rise sierras and isolated peaks of volcanic origin,
the highest in North America, their summits covered with eternal
snow, which shelter, temper, and protect the fertile plateaux lying at
their base. Centrally located on this table-land, surrounded by a wall
of lofty volcanic cliffs and peaks, is the most famous of all the valley
plateaux, something more than one hundred and sixty miles in
circuit, the valley of Mexico, Anáhuac, that is to say, 'country by the
waters,' taking its name from the lakes that formerly occupied one
tenth of its area. Anáhuac, with an elevation of 7,500 feet, may be
taken as representative of the tierra fria. It has a mean temperature
of 62°, a climate much like that of southern Europe, although dryer,
and to which the term 'cold' can only be comparatively applied. The
soil is fertile and productive, though now generally presenting a bare
and parched surface, by reason of the excessive evaporation on lofty
plains exposed to the full force of a tropical sun, its natural forest-
covering having been removed since the Spanish conquest, chiefly, it
is believed, through artificial agencies. Oak and pine are prominent
features of the native forest-growth, while wheat, barley, and all the

European cereals and fruits flourish side by side with plantations of
the indigenous maize, maguey, and cactus. From May to October of
each year, corresponding nearly with the hot season of the coast,
rains or showers are frequent, but rarely occur during the remaining
months. Trees retain their foliage for ten months in the year, and
indeed their fading is scarcely noticeable. Southward of 18°, as the
continent narrows, this eastern table-land contracts into a mountain
range proper, presenting a succession of smaller terraces, valleys,
and sierras, in place of the broader plateaux of the region about
Anáhuac. Trending south-eastward toward the Pacific, and uniting
with the western Sierra Madre, the chain crosses the isthmus of
Tehuantepec at a diminished altitude, only to rise again and expand
laterally into the lofty Guatemalan ranges which stretch still south-
eastward to Lake Nicaragua, where for the second time a break
occurs in the continental cordillera at the southern limit of the
territory now under consideration. From this central cordillera lateral
subordinate branches jut out at right angles north and south toward
either ocean. As we go southward the vegetation becomes more
dense, and the temperature higher at equal altitudes, but the same
gradations of 'fria' and 'templada' are continued, blending into each
other at a height of 5,000 to 6,000 feet. The characteristics of the
cordillera south of the Mexican table-land are lofty volcanic peaks
whose lower bases are clothed with dense forests, fertile plateaux
bounded by precipitous cliffs, vertical fissures or ravines of immense
depth torn in the solid rock by volcanic action, and mountain
torrents flowing in deep beds of porphyry and forming picturesque
lakes in the lower valleys. Indeed, in Guatemala, where more than
twenty volcanoes are in active operation, all these characteristic
features appear to unite in their highest degree of perfection. One of
the lateral ranges extends north-eastward from the continental
chain, forming with a comparatively slight elevation the back-bone of
the peninsula of Yucatan.
THE TIERRA CALIENTE.

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com

Advances In Computer Graphics 24th Computer Graphics International Conference Cgi 2006 Hangzhou China June 2628 2006 Proceedings 1st Edition Monssef Alsweis

About This Presentation

Slide Content

Tags

Categories

Download

Quick Actions

Statistics

Related Slideshows

Advances In Computer Graphics 24th Computer Graphics International Conference Cgi 2006 Hangzhou China June 2628 2006 Proceedings 1st Edition Monssef Alsweis

About This Presentation

Slide Content

Slide 1

Slide 2

Slide 3

Slide 5

Slide 6

Slide 7

Slide 8

Slide 9

Slide 10

Slide 11

Slide 12

Slide 13

Slide 14

Slide 15

Slide 16

Slide 17

Slide 18

Slide 19

Slide 20

Slide 21

Slide 22

Slide 23

Slide 24

Slide 25

Slide 26

Slide 27

Slide 28

Slide 29

Slide 30

Slide 31

Slide 32

Slide 33

Slide 34

Slide 35

Slide 36

Slide 37

Slide 38

Slide 39

Slide 40

Slide 41

Slide 42

Slide 43

Slide 44

Slide 45

Slide 46

Slide 47

Slide 48

Slide 49

Slide 50

Slide 51

Slide 52

Slide 53

Slide 54

Slide 55

Slide 56

Slide 57

Slide 58

Slide 59

Slide 60

Slide 61

Slide 62

Slide 63

Slide 64

Slide 65

Slide 66

Slide 67

Slide 68

Slide 69

Slide 70

Slide 71

Slide 72

Slide 73

Slide 74

Slide 75

Slide 76

Slide 77

Slide 78