3D Sampling Performative Textures Remixed from 3D Scans

3dj: 3d Sampling Hapticand Optically Performative TexturesRemixed from 3d Scans

Sayjel V. Patel and Caitlin T. Mueller

Abstract3D Sampling is presented as a new method for comprehensiverearrangement and composition of 3D scan data, allowing multi-scalemanipulation of surface texturing and subsequent fabrication andre-evaluation using 3D printing. Analogous to sample-based music, 3DSampling contributes new frameworks, tools, and compositional strategiesfor the digital remixing, hacking, and appropriating of material qualities,performances, and behaviours directly from physical samples to producenew designs. Case studies are presented which demonstrate 3DJ: aprototype 3D modelling tool for synthesizing haptic and opticallyperforming textures from 3D scan-generated source material, which canbe applied in the context of other 3D modelling techniques.

Introduction

3D Sampling is part of a revolution in sensibili-ties in the field of design modelling. For the firsttime, hand-held 3D scanning technologies, suchas photogrammetry, enable anyone with asmart-phone to digitally capture real worldobjects as point-cloud data. In the near future, atreasure trove of 3D materials sampled from realworld objects and environments will be availableto designers. Presently, the rules of computer

modelling software precede and constrain theform of the objects generated. However, 3Dscans allow a designer to reverse this methodol-ogy, making detailed digital replicas of physicalobjects the starting point for design (Fig. 1). Bycontributing methods to interject and collide thecomplex or structural irregular geometries of 3Dscans deliberately into playfield of digital design,3D Sampling facilitates a new solution space thatwould otherwise be inconceivable through con-ventional optimized, rule-based, parametric, orprocedural modelling practices. In contrast withapproaches taking inspiration from observedphenomena to infer design strategies, such asbio-mimicry and material computation (Oxman2009), 3D Sampling allows designers to appro-priate, hack, and remix the material behaviour ofmacro-scale texture geometry such as structuralcorrugation, water shedding, grip, acoustics,

S.V. Patel (&)Massachusetts Institute of Technology, MIT-SUTDCollaboration, Cambridge, USAe-mail: [email protected]

C.T. MuellerMassachusetts Institute of Technology, Departmentof Architecture, Cambridge, USA

© Springer International Publishing Switzerland 2015M.R. Thomsen et al. (eds.), Modelling Behaviour, DOI 10.1007/978-3-319-24208-8_44

527

optics, and camouflage (Gage 2013). This paperpresents the conceptual framework and techno-logical implementation of 3D Sampling, dem-onstrated through case studies presenting designsfor architectural surfaces textures.

Related Work

Akin to sample-based music, 3D Sampling com-bines existing modelling tools and methods withconcepts and practices drawn from externalexamples. This section describes: (1) How sam-pling provides the conceptual inspiration forapplying 3D scanners as design generators;(2) Recent work in 3D mesh sculpting and cut andpaste interfaces; (3) The technological opportu-nities and challenges related to 3D scanning;(4) Research efforts in the computer graphicscommunity exploring texture synthesis tech-niques; and (5) The functional and cultural sig-nificance of texture design in contemporaryarchitectural practice.

Sampling

In music, sampling is the act of taking a portionof a sound or recording and reusing it to createoriginal compositions. Walker (2008) provides acomprehensive overview of the field ofsample-based music; its motivations, composi-tional strategies, and history. Some of the central

manipulations applied by artists in this field,including digital signal processing, convolution,algorithmic filtering, looping, and mixing, serveas compelling metaphors for 3D Sampling. Inarchitecture, 3D scanning (Sheil 2013) and tex-tural sampling (Gage et al. 2013) have garneredrecent attention; however, hybrid applications, asproposed in this paper are scarce.

Personal 3D Scanning

3D scanners are instruments that collect data aboutthe geometry of physical objects, outputting listsof geo-spatial coordinates called point-clouds.Photogrammetry is one such 3D scanning tech-nology that generates textured 3D meshes fromphotographs. However, because of computationalrequirements, there is a significant time lagbetween scanning and model generation. In con-trast, low-cost, real-time scanning can be achievedthrough depth camera technology. KinectFusionapplies this technology to ‘paint’ 3D reconstruc-tions through the motion of Kinect device (Azadiet al. 2011). One inherent disadvantage of thistechnology is the trade-off between the resolutionof reconstructions and the immediacy of theinteraction. As discussed in the literature (Bou-lassai et al. 2009; Arachchige and Perera 2014;Truong-Hong and Laefer 2011), a fast and reliablemethod for real-time segmentation and conversionof point-clouds into discrete 3D surfaces forcomputer modelling remains in absentia.

Fig. 1 3D Sampling overview: a Photograph, rockformation with convex surface features. b Detail, photo-grammetric model generated using AutoDesk 123d Catch

(http://www.123dapp.com/catch). c 3DJ: prototype designtool for textured synthesis from 3D scan samples. d 3Dprinted texture sample

528 S.V. Patel and C.T. Mueller

Procedural Texture Generation

Image-based texture synthesis is a topic of con-tinued interest in the field of computer graphics,whereby 2D samples from photographic imagescan be used to synthesize entire worlds (Efrosand Freeman 2001). Texture synthesis algorithmshave two primary goals: to sample an existingtexture to generate an unlimited amount of newimage data that perceived to be the same tohumans, and to transfer textures between objects.Texture synthesis works by extracting patches ofan existing texture and applying the samples inconsistent way. The literature provides a range ofpossible technical approaches to 2D texturalsynthesis (Ashikhmin 2001; Ruiters et al. 2010;Efros and Leung 1999; Wei and Levoy 2000).While typical applications of textural synthesis incomputer graphics are guided byperformance-driven criteria, 3D sampling appliestextural synthesis as a new technique for designexploration.

Virtual Clay

Evocative of traditional practices of physical claymodelling, virtual clay has garnered recentattention as an alternative to conventional 3Dmodelling styles. A review of scholarly workreveals 3D Sculpting (Dipen et al. 2013; Cit et al.2013; Cingi and Oghan 2012; Madugalla et al.2013) and Cut and Paste modelling (Sharf et al.2006; Schmidt and Singh 2010; Bierman et al.2002) as two inter-related applications of virtualclay. This body of work has leveraged a varietyof rapid mesh editing techniques to produceinnovative new compositional operations forfusion of discrete parts and the transfer of surfacefeatures (Schmidt and Singh 2010). Bringingtogether the processes of concept visualizationand final design virtual clay offers a way toquickly explore a large space of alternativesolutions by extending the functionality ofexisting 3D objects and simplifying user inter-action and manual manipulation (Sharf et al.2006). When used in conjunction with virtual

clay, 3D Sampling contributes novel remixingoperations which allow a designer extend andexploit a new range possibilities beyond the meresurface of an input 3D objects.

Textures, Patterns, Behaviours

Architects have always relied on textures andornamentation as a compelling way of resolvingthe cultural and pragmatic requirements ofbuilding details (Fig. 4a). Considering digitaltechnologies and networks, the status of archi-tectural texture and ornament has shifted inresponse to parametric tools, (e.g. Grasshopperor Maya); now designers can create highlycomplex and dynamic patterns predicated onrelatively simple rules. The emerging interest intexture and ornament is discussed extensively inthe literature (Moussavi and Kubo 2006; Lynn2004; Picon 2013; Gleiniger and Vrachliotis2009; Levit 2008).

Research Question

The framework and tools presented here addresssome of the major limitations inherent in cur-rently available computer modelling approaches—specifically, how to define the rules of aparametric system which produce a desired formor material behaviour. Without a clear under-standing of the rules from the onset, parametricand procedural modelling approaches ineffi-ciently reach acceptable solutions requiring sig-nificant iteration through trial and error.Demonstrating a new field of design and researchinquiry, this paper explores several broadlydefined questions: What advantages does 3DSampling offer as a technique for design explo-ration? What are some possible applications?What is the relationship (if any) between thesampled material and the subsequent textures?How does one evaluate or analyze results? Whatare the boundaries of such an approach? Wepostulate that it is possible to 3D scan specific

3dj: 3d Sampling Haptic and Optically Performative … 529

qualities of an object, remix the material digi-tally, and produce a range of new materialqualities from the original scan (Fig. 2).

Methodology

This section outlines the details of 3D Sampling,a novel technique for synthesizing textures withmaterial qualities one or more 3D scans as thefoundation. The function and rationale of thenew 3DJ 3D Sampling software tool iselucidated.

Overview

Figure 3 outlines the six operations 3D Sam-pling contributes to traditional 3D modelling.Figure 4 explains conceptual framework of 3DSampling. The key processes described are:(1) Scanning—the digitization of a physicalsample. (2) Synthesis—remixing a digital sam-ple using conventional (e.g. Rhino) and novel(3DJ) digital modelling. (3) Printing—physicalmaterialization of a digital sample, and(4) Analysis—evaluating the behaviour or per-formance of a physical or digital sample. Theframework also tracks the format of informationas it translates mediums throughout the sam-pling process: (1) physical samples—physicalobjects; point-cloud samples—non-hierarchicallists of coordinates, and 3D mesh samples—hierarchical lists of coordinates.

Technical Frameworkand Implementation

This section describes the technical implemen-tation of 3D sampling further.

3D Synthesis

3D Synthesis involves the processes of sampling,modifying, and remixing digital samples usingconventional and novel software approaches.This section describes 3DJ, a custom DJ tooldeveloped for manipulating 3D scan point-cloudsin the course of this research.

3DJ

Inspired by music sampling tools and methods,3DJ is a prototype computer modelling applica-tion developed by the authors for 3D Sampling(Figs. 5, 6 and 7). Written in Processing, theinputs and outputs of 3DJ are point-clouds. 3DJprovides an interface and additional featuresallowing designers to manipulate point-clouds innew ways. 3DJ recognizes the practices, meth-ods, and procedures of design established insample-based music, and organizes the experi-ence of designing with 3D scans into six opera-tions—sample, cut, mix, filter, analyze, and loop—to produce new texture variations The 3DJinterface allows a user to crop regions from

Fig. 2 Performative textures: a Macabres, Alhambra(source http://en.wikipedia.org/wiki/Alhambra#/media/File:Ceiling_in_Alhambra.JPG, user Liam 987). b Haptic

texture study, a range of surface textures were generatedusing the 3DJ modelling tool and CNC milled into a slabof sapele wood


Fig. 3 Summary of 3DJ operations and functions: (1)Sample—a set of point cloud data, (2) Cut—extracting aset of point cloud data, (3) Loop—repetition of geometricfeatures, using a process of arraying tiles (3.1) or pixelquilting (3.2) as described in (Efros and Freeman 2001),(4) Mix—textural blending, superposition, and juxtaposi-tion. Techniques include superposition and averaging ofsamples (4.1), linear blending using distributed averagingof pixel positions (4.2), transforms using image analogies

(Hertzmann et al. 2001), and 3D mapping of texturesusing techniques such as projective texture mapping(Segal et al. 1992), (5) Filter—including, mesh reductionthrough localized averaging of clusters (5.1), meshsmoothing, coordinate transforms, particularly useful asa precursor to mapping operations (5.2), (6) Analysis—evaluation and identification of desirable features of asample, either through empirical testing methods (6.1), orcomputational simulation (6.2)


Fig. 4 3D Sampling network diagram: flows, processes and tools


imported 3D scan point-clouds or meshes (cut),and apply looping, mixing, or filtering operationsto synthesize a new point-cloud texture whichcan is output as a PLY file. While more sophis-ticated textural synthesis algorithms can beincorporated in the future, the ‘looping’ opera-tion involves a simple mirroring and tiling ofsub-samples. ‘Mixing’ involves a simple linearblending algorithm to combine two or morescans (Fig. 8). The ‘Filter’ in this version of 3DJ

involves operations to apply global mathematicaldistortions across the textural field, as well asmesh reduction algorithms; either involving arandom re-sampling of the input point cloudwithin 3DJ, as well as the use of the built-inmesh reduction function in rhino prior to theimport of a sample being imported into3DJ. While future versions of 3DJ could includereal-time mesh re-construction; efforts to do sousing within 3DJ stalled the tool, undermining

Fig. 5 Screenshot of 3DJ interface: a 3D view port. b 2Dinterface, here a user crops of a sub-sample from the input3D scan data. c Customizable buttons for user manipu-lation, toggles a looping operation, mathematical filters,

point cloud reduction and coordinate mapping. d 3D viewselector—top, bottom, left, right, axo—southwest, south-east, northwest, northeast

Fig. 6 Looping operation applied to sub-sample cropped from a 3D scan


Fig. 7 Coordinate transformations (clockwise starting from top left): x-y-z texture, cylindrical coordinate system, polarcoordinates, spherical coordinates

Fig. 8 Linear blending—contour drawing of atexture combining four 3Dsampled textures


interaction rate; ultimately a Poisson surfacereconstruction algorithm and the built-in Dela-unay mesh triangulation were applied usingMesh Lab to produce satisfactory meshes forsubsequent milling and 3D printing operations.While similar techniques could be achieved byemploying a variety of existing software inconcert, the aim of integrating these operationswithin a single tool was to facilitate greaterdesign exploration (Fig. 9) by allowing the userto actively intervene within the process of guidedtextural synthesis.

Analysis

Current technological limitations constrain 3Dsampling to macro-scale material properties,attributes and characteristics. For the initial tests,high definition starch-based powder 3D printing(Fig. 10a) was used. However the brittle, watersoluble, nature of this medium degraded perfor-mative qualities, thus other alternatives wereexplored. 3D printing mediums like resin-basedstereo lithography (Fig. 10b) and ABS allowoptically and mechanically performative parts.

Fig. 9 Design exploration: variations created fromprototype 3D texture design tool: a Sample: a croppedextract of an oyster shell. b Filter: a mesh-reduce

operation and mesh smoothing operation. c Mix: amathematical z-distortions applied to textures. d Loop:tiling of sample textures achieves scalar-manipulations

Fig. 10 a Additive manufacturing: powder printed models were brittle and did not have useful mechanical properties.b Additive manufacturing optical qualities—resin stereo lithography on a form-1 3D printer. c CNC milling


However, economies of scale and material flex-ibility made CNC milling the desired fabricationmedium here.

Analogous to how musicians must have theability to listen to a progressing mix to properlyjudge their compositions, 3D Sampling requiresfeedback to evaluate how mixing operationsmanifest as the sample’s physical characteristics.While the initial test applied qualitative analysisand evaluation in the production of texturedesign (Fig. 11), other possible modes ofempirical, and real-time digital analysis exist,including thermal, optical, and haptic perfor-mances, and tests quantifying surface qualities(e.g. aerodynamics, roughness, irradiance, etc.)While the physical analysis methods provideimmediate, tangible results and verification,real-time digital analysis methods within 3DJ

would facilitate a stronger relationship betweendigital manipulations the final texture designs.

Case Studies

This section describes initial design experimentsdemonstrating the application of the 3D samplingframework to the design of camouflage, haptic,and optical, qualities using a single 3D scan.

Landscape and Camouflage Textures

Figures 12, 13 and 14 describe the first designexperiment. Imagining possible applications tolandscape design, this test demonstrates how

Fig. 11 Initially analysis, exploring how texture designs dispersed fluid across a surface

Fig. 12 Landscape scale

Fig. 13 Proto-architectural scale


terrestrial features can be 3D scanned and blen-ded at multiple scales of occupation, generating alandscape. Textures representing a range ofscales using sampling and looping operationswere synthesized in 3DJ and combined through atechnique of superposition, to produce a hypo-thetical landscape providing numerous scales ofhuman habitation. Considering the human bodyas a datum, we observe three scales in the model:A landscape scale, suggestive of a connection toa broader context, a proto-architectural scale,suggestive of more enclosed and intimate spaces,and finally a human-scale, suggestive of ways inwhich a person might directly engage thelandscape.

Haptic Textures

The second series of experiments (Figs. 15 and16) explored applications at product design scaleby demonstrating possibilities to synthesize tex-tures with tactile qualities (e.g. grip, comfortetc.). Multiple sampling, looping, mixing, andfiltering operations were applied using 3DJ to

generate a wide range of material textures: fromslippery, to sharp, to rough. A key findingrevealed that the material selection itself was adesign operation, transforming the qualities ofthe raw 3D scan. For example, an otherwiseun-altered geometry sampled from the 3D scanbecomes smooth like a piece of cloth whenmilled into wood. However, applying a filterproduces a jagged texture. Applying a loopincreases the roughness of the sample. Oneopportunity demonstrated in this case study isusing 3D Sampling as a means to short-scalesurface heterogeneity (Fig. 17).

Fig. 14 Human-scale

Fig. 15 Material selection and the translation of material qualities (touch) from the original sample

Fig. 16 Haptic case study: qualitative evaluation


Optical Textures

Figures 18, 19 and 20 highlight the third exper-iment, exploring the synthesis of optically per-formative textures at the building envelope scale.The methods were applied to demonstrate howmaterial qualities related to light transmissionacross a CNC milled acrylic surface such asopacity, transparency, and reflectivity could becontrolled using the 3DJ tool. From the pre-liminary results, it was observed that filtering,which removes redundant points, producesgreater transparency (Fig. 17), while looping,which increases point density creates a texturewith greater diffusion and specular noise(Fig. 18). The unaltered sample geometry(Fig. 19), results in textural chimera.

Fig. 17 Textural gradients

Fig. 18 Optical example1, mesh reduction filterachieves more directedlight

Fig. 19 Looping textureproduces more dispersedlight and opaqueness

Fig. 20 Unalteredgeometry texture from theoriginal 3D scan


Case Study Summary

The motivation for the case studies was todemonstrate morphogenic potentials of sampledmaterial. Part of this involved the development ofsyntax to link operations of the tool and theresultant material performances and behavioursproduced. Figures 21, 22 and 23 summarize theapplication of 3D Sampling operations, andcompositional strategies used to generate per-formative textures. The case studies provide ini-tial results, demonstrating a single iteration of the3D Sampling framework.

Conclusion

Summary

3D Sampling imagines a future where a designercould instantly sample and remix the geometryand material qualities of objects from the realworld, producing new designs. The research

presented in this paper contributes: (1) The con-cept of 3D Sampling as a creative process; (2) Adefinition for conceptual and technical frameworkfor 3D Sampling; (3) 3DJ, a 3D modelling soft-ware tool allowing users to generate new 3Dtextures from 3D scans; and (4) Three casestudies implementing the framework, initialresults and evaluation. The speculative tools,frameworks, and experiments presented andassessed in the paper, represent first steps whichexemplify the potential future of a new design andresearch field which re-contextualizes 3D scan-ning, once conceived as a process of measure-ment, as a method for creative production.

Potential Impact

While the solutions developed within the threecase studies—spatial, haptic, and optical—wereframed within an architectural discourse of tex-ture, the implications of this work reach faroutside the discipline. Considering how

Fig. 21 Camouflage case study

Fig. 22 Haptic case study

Fig. 23 Optical case study


sample-based music practices have made musiccreation accessible to a wider audience, 3Dsampling could potentially offer a simplified andmore intuitive way for non-experts to edit,manipulate, and create 3D objects. Future 3Dtextural synthesis tools, suggested by the 3DJprototype software, imply a wide range of pos-sible applications in light of different users. Forinstance a product designer could apply thesemethods to design ergonomic, tactile, qualitiessuch as grip, comfort, water shedding, around thesense of touch, while a landscape or urbandesigner might sample of natural features tointegrate a structure into a landscape throughcamouflage. Conversely, a materials engineercould use 3D Sampling to apply a texture withdesirable properties but poorly understood prin-ciples such as gecko’s feet to a develop andanalyze solutions directly through empiricallytesting, contrasting to parametric modelling pro-cesses which require the development of a con-ceptual model of a gecko’s foot prior toapplication within a design space.

Future Work

To further explore the connection between theoperations and performance, future work mayincorporate additional iterations between syn-thesis and analysis in order to further knowledge.Quantitative analysis and experimental datawould buttress this. Greater interrogation andclassification of the initial samples, as well thecombination of multiple samples would expandon this initial exploration, such as the possibilityof creating material hybrids (e.g. mixing gecko’sfeet with fur). A promising avenue to advancemethods regarding mixing within 3DJ would beto adapt seminal work on 2D texture synthesis incomputer graphics (Efros and Freeman 2001;Ruiters et al. 2011; Hertzmann et al. 2001) as away to conceive of new mixing, looping, andfiltering operations. Bump mapping techniquescould provide a quick way to test how thesetechniques for 2D textural synthesis translate to3D point-clouds. Another possible topic for

exploration is the mapping process to 3D texturesto be applied irregular surfaces, and their rela-tionship to surface performance.

Concluding Remarks

Within architectural practice, the use of externaldesign precedents, research, and first-handobservations as forms of knowledge are oftenconsidered more efficient strategies in developingdesigns than initiating projects from a tabula rasa(Zarzar 2003). 3D scanning generates a wealth ofnew digital data that architects can access freelyto produce new material, spatial, and even cul-tural qualities, properties, and effects. Byre-framing 3D scanning as a mode of creativeproduction and operationalizing this modethrough novel software demonstrations, thispaper explores the ability to transcend subjectiveor experiential qualities of physical reality, andempowers designers with new tools to experi-ence and remix the world.

Acknowledgments Special thanks to: Skylar Tibbits,Terry Knight, Marc Downie, and Mark Goulthorpe fortheir mentorship, timely ideas, and guidance, Inés Ariza,for being an excellent colleague, and her generous con-tribution towards the photographic documentation pre-sented here, Alexander Charidis, for sharing his insightsand knowledge of related research efforts in the field ofcomputer graphics, Ben Goldstein, for his help editingand clarifying earlier drafts of this paper, and theMIT-SUTD Collaboration and the MIT Structural DesignLab for supporting the work.

References

Ashikhmin M (2001) Synthesizing natural textures. In:Proceedings of the 2001 symposium on interactive 3Dgraphics—SI3D ‘01

Biermann H, Martin I, Bernardini F, Zorin D (2002)Cut-and-paste editing of multiresolution surfaces.ACM Trans Graph 21(3):312–321

Boulaassal G, Landes T (2009) Automatic extraction ofplanar clusters and their contours on building facadesrecorded by terrestrial laser scanner. Int J ArchitComput 07(1):1–20

Cingi C, Oghan F (2012) Teaching 3D sculpting to facialplastic surgeons. Facil Plastic Surg Clin Am 19(4):603–616


Efros AA, Leung TK (1999) Texture synthesis bynon-parametric sampling. Proc Seventh IEEE Int ConfComput Vis 2:1033–1038

Efros AA, Freeman WT (2001) Image quilting for texturesynthesis and transfer. In: Proceedings of the 28thannual conference on computer graphics and interac-tive techniques—SIGGRAPH ‘01

Gage M, Friddle T, Ayas T, Burry M, Cannavino B,Connolly R, Kehagias N, McPhail L, Oncescu C,Sheridan W, Stranix K, Tripodi RJ, Utting B, Wis-kup E (2013) Disheveled geometries: the digital stoneproject. Yale School of Architecture, New Haven

Gleiniger A, Vrachliotis G (2009) Pattern: ornament,structure, and behavior. Birkhäuser, Basel

Hertzmann A, Jacobs CE, Oliver N, Curless, Salesin DH(2001) Image Analogies. In: Proceedings of the 28thannual conference on computer graphics and interac-tive techniques—SIGGRAPH ‘01

Izadi S, Davison A, Fitzgibbon A, Kim D, Hilliges O,Molyneaux D, Newcombe R, Kohli P, Shotton J,Hodges S, Freeman D (2011) KinectFusion. In:Proceedings of the 24th annual ACM symposium onuser interface software and technology—UIST ‘11

Levit R (2008) Contemporary “ornament”: the return ofthe symbolic repressed. Harvard Des Mag 28:70–85

Lynn G (2004) The structure of ornament. In: Leach N,Turnbull D, Williams C (eds) Digital tectonics.Wiley-Academy, Chichester, pp 62–68

Madugalla AK (2013) FaceID: a 3D computer graphicapplication for forensic medicine: a novelsemi-automated muscle based digital sculpting initia-tive for forensic facial reconstruction in Sri Lanka. In:2013 International conference on computer medicalapplications (ICCMA). pp 1–6

Moussavi F, Kubo M (2006) The function of ornament.Actar, Barcelona

Menges A (2012) Material computation: higher integra-tion in morphogenetic design. Wiley, Hoboken

Oxman N, Rosenberg JL (2007) Material-based designcomputation: an inquiry into digital simulation ofphysical material properties as design generators. Int JArchit Comput 5(1):26–44

Segal M, Korobkin C, Van Widenfelt R, Foran JJ,Haeberlie P (1992) Fast shadows and lighting effectsusing texture mapping. In: Proceedings of SIG-GRAPH ‘92. pp 249–52

Schmidt R, Singh K (2010) Meshmixer. In: ACM SIG-GRAPH 2010 Talks on—SIGGRAPH ‘10

Sharf A, Blumenkrants M, Shamir A, Cohen-Or D (2006)SnapPaste: an interactive technique for easy meshcomposition. Vis Comput Vis Comput 22(9–11):835–844

Ruiters R, Schnabel R, Klein R (2010) Patch-basedtexture interpolation. Comput Graph Forum 29(4):1421–1429

Picon A (2013) Ornament: the politics of architecture andsubjectivity. Wiley, Hoboken

Miller PD (2008) Sound unbound: sampling digital musicand culture. MIT, Cambridge

Wei LY, Levoy M (2000) Fast texture synthesis usingtree-structured vector quantization. In: Proceedings ofthe 27th annual conference on computer graphics andinteractive techniques—SIGGRAPH ‘00

Zarzar KM (2003) Use and adaptation of precedents inarchitectural design: toward an evolutionary designmodel: proeschrift


3D Sampling Performative Textures Remixed from 3D Scans

Documents

Transcript of 3D Sampling Performative Textures Remixed from 3D Scans