Downloaded from orbit.dtu.dk on: Feb 06, 2023 Light, Matter, and Geometry The Cornerstones of Appearance Modelling Frisvad, Jeppe Revall Publication date: 2008 Document Version Publisher's PDF, also known as Version of record Link back to DTU Orbit Citation (APA): Frisvad, J. R. (2008). Light, Matter, and Geometry: The Cornerstones of Appearance Modelling. DTU Compute PHD General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Users may download and print one copy of any publication from the public portal for the purpose of private study or research. You may not further distribute the material or use it for any profit-making activity or commercial gain You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Light, Matter, and Geometry The Cornerstones of Appearance Modelling Jeppe Eliot Revall Frisvad Kongens Lyngby 2008 IMM-PHD-2008-188 Technical University of Denmark Informatics and Mathematical Modelling Building 321, DK-2800 Kongens Lyngby, Denmark Phone +45 45253351, Fax +45 45882673 [email protected] www.imm.dtu.dk IMM-PHD: ISSN 0909-3192 Abstract This thesis is about physically-based modelling of the appearance of materials. Whenamaterialisgraphicallyrendered,itsappearanceiscomputedbyconsid- ering the interaction of light and matter at a macroscopic level. In particular, theshapeandthemacroscopicopticalpropertiesofthematerialdeterminehow itwillinteractwithincidentillumination. Inthisthesisthemacroscopicoptical properties are connected to the microscopic physical theories of light and mat- ter. This enables prediction of the macroscopic optical properties of materials, and, consequently, also prediction of appearance based on the contents and the physical conditions of the materials. Physically-based appearance models have many potential input and output pa- rameters. There are many choices that must be made: How many material components to include in the model, how many physical conditions to take into account,whethertheshapeofthematerialshouldbecoupledtotheappearance model or not, etc. A generalised concept of shape and geometry is presented to provide a framework for handling these many degrees of freedom. Constraints betweeninputandoutputparametersaremodelledasmultidimensionalshapes. Thisgivestheopportunitytousetheappearancemodelsnotonlyforprediction, but also for analysis of the contents and the physical conditions of a material given information about its macroscopic optical properties. Since it is possible to measure these properties using camera technology, the presented framework enables analysis of material contents and conditions using camera technology. Three detailed appearance models are presented as to exemplify the applica- bility of the theory: (1) A model which finds the appearance of water given temperature, salinity, and mineral and algal contents of the water; (2) a model which finds the appearance of ice given temperature, salinity, density, and min- eral and algal contents of the ice; and (3) a model which finds the appearance of milk given fat and protein contents of the milk. ii Resum´e Denneafhandlingomhandlerfysiskbaseretmodelleringafmaterialersudseende. N˚aretmaterialebliverfremstilletgrafisk,beregnesdetsudseendevedatoverveje vekselvirkningen mellem lys og substans p˚a et makroskopisk niveau. Mere præcist bestemmer materialets form og dets makroskopiske optiske egenskaber, hvordan det vil vekselvirke med indfaldende belysning. I denne afhandling kop- les de makroskopiske optiske egenskaber til de mikroskopiske fysiske teorier for lys og substans. Dette muliggører forudsigelse af materialers makroskopiske optiske egenskaber, og derfor ogs˚a forudsigelse af deres udseende v.h.a. materi- alernes indhold og deres fysiske tilstande. Fysisk baserede modeller for udseende har mange potentielle input og output parametre. Derermangevalg,somm˚agøres: Hvormangematerialekomponen- ter der skal inkluderes i modellen, hvor mange fysiske tilstande der skal tages højde for, om materialets form skal koples til modellen for udseendet eller ej, o.s.v. Etgenereliseretkonceptm.h.t.formoggeometripræsenteresforatgiveet system til behandling af disse mange frihedsgrader. Bindinger mellem et mate- rialesinputogoutputparametremodelleressomenmultidimensionelform. Det givermulighedforatbrugemodellerforudseendeikkekuntilforudsigelse,men ogs˚a til analyse af et materiales indhold og dets fysiske tilstande givet informa- tion om dets makroskopiske optiske egenskaber. Da det er muligt at m˚ale disse egenskaber ved brug af kamerateknologi, muliggører det præsenterede system ogs˚a analyse af materialeindhold og tilstande ved brug af kamerateknologi. Der præsenteres tre detaljerede modeller for udseende for at give eksempler p˚a teoriens anvendelsesmuligheder: (1) En model til at finde vands udseende givet vandets temperatur, dets saltindhold og dets indhold af mineraler og alger; (2) enmodeltilatfindeis’udseendegivetisenstemperatur,saltindhold,massefylde ogdensindholdafmineralerogalger;og(3)enmodeltilatfindemælksudseende givet mælkens protein- og fedtindhold. iv Preface Study without thinking, and you are blind; think without studying, and you are in danger. Confusius (552 B.C. – 479 B.C.), from the Analects (2:16) This thesis was written out of curiosity and fascination with nature. It was prepared at the department of Informatics and Mathematical Modelling of the Technical University of Denmark in partial fulfillment of the requirements for acquiring the Ph.D. degree in mathematical modelling. The subject of the thesis is appearance modelling which is a subject within the branchofcomputergraphicsknownasrealisticimagesynthesis. Thestudentof realisticimagesynthesisisprivilegedinbeingallowedtoinvestigatethereasons forallvisualaspectsofnature. Indeedallvisualaspectsareinteresting,andthe main objective is to capture their appearance correctly. It seems that this ob- jective is most sensibly attained by physical models. However, the gap between thephysicist’sunderstandingofnatureatthemicroscopiclevelofquantumpar- ticles and the modelling of appearance is wide. It is the aim of this thesis to buildabridgeoverthegap,orperhapsjusttolaythefoundationforthebridge. This is done in three parts: One part concerning light, one concerning matter, and one concerning geometry. A part on each of the three cornerstones of ap- pearance modelling. Finally, there is a fourth part in which appearance models are developed based on the theory provided in the first three parts. Itisunusualforathesistospreadoversuchalargevarietyoftheoriesasyouwill find in this one. The usual approach would be to focus as narrowly as possible on the perfection of a single technique. The reason for the unusual approach is the gap and the missing bridge. If there had been a book closing the gap, or providing the bridge, this thesis would have been entirely different. As it is, vi there is no such book. So instead of building a house with no foundation, this thesis became an attempt on laying the foundation itself. Ingraphicswewouldliketobeabletomodeltheappearanceofasmanydifferent materials as possible. Therefore the theory has been kept as general and as flexible as possible throughout the thesis. It is in the effort to do so that the main contributions of the thesis appear. This makes it difficult to split the thesis in a part on background theory and a part on contributions. Instead, the introduction contains a relatively detailed overview of the thesis in which the contributions are pointed out. Thereaderisassumedtohavesomegraduate-levelmathematicalunderstanding and some general knowledge about graphics, in particular ray tracing. The project was advised by Associate Professor Niels Jørgen Christensen and Professor Peter Falster who are both with the department of Informatics and Mathematical Modelling of the Technical University of Denmark. The project has been funded by a Ph.D. scholarship from the Technical University of Den- mark. Part of the work presented in this thesis has been published, or is to appear. Details on these publications are provided in the Acknowledgements. Thehumanmindisafantasticimageprocessor. Whenweobservenature,weare able to draw conclusions based on subtle details. Especially if we know what to look for. Mathematical models which describe the appearance of nature have the ability to teach us what to look for. They can tell us the visual consequenceofchangingthecontentsofamaterial,orchangingthetemperature, or changing other properties. In this way, appearance models make observation of nature more instructive, but also more spectacular. Let us build the bridge and construct more models. Lundtofte, November 2007 Jeppe Revall Frisvad Acknowledgements FirstandforemostthankstoMonicaforherloveandsupport,andforhercoping withmyabsenceatmanylonglatehours. Alsotomyparents,SvenandFelicia, for their everlasting support, and for teaching me to choose my education out of interest. To my brother Rasmus, thanks for our shared effort in learning about graphics during our Master’s studies, and to him, and to Ulrik Lund Olsen, thanks for many good suggestions as to improvements of this thesis. Thanks to my Ph.D. advisers Niels Jørgen Christensen and Peter Falster for their help in getting a Ph.D. scholarship for me, for their unfailing support throughout my studies, and for their mild guidance such that I was always able to pursue my own ideas. A special thanks to Andreas Bærentzen for his open door policy, and for our numerous discussions on many aspects of the work presented in this thesis (I hope that I did not take too much of your time). Iwouldliketothanktheadvisersofmyexternalstays. HenrikWannJensenfor suggesting that we write a SIGGRAPH paper based on my work, and for the great help and hospitality given to me by him and his family during my stay in San Diego in January 2007. Geoff Wyvill for sharing his profound knowledge about noise and geometry, and for welcoming my three months visit at the University of Otago, New Zealand, in 2006. Thanks also to Gert L. Møller for taking an interest in my work, and for co- authoring some of the first papers published during this project. Other people who have had a positive influence on my work, and who I would
Description: