Data to Physicalization: A Survey of the Physical Rendering Process
Hessam Djavaherpour, Faramarz Samavati, Ali Mahdavi-Amiri, Fatemeh Yazdanbakhsh, Samuel Huron, Richard Levy, Yvonne Jansen, Lora Oehlberg
EEUROVIS 2021N. Smit, K. Vrotsou, and B. Wang(Guest Editors)
Volume 40 ( ), Number 3STAR – State of The Art Report
Data to Physicalization: A Survey of the Physical Rendering Process
H. Djavaherpour , F. Samavati , A. Mahdavi-Amiri , F. Yazdanbakhsh , S. Huron , R. Levy , Y. Jansen , and L. Oehlberg University of Calgary, Simon Fraser University, Institut Polytechnique de Paris, CNRS. Sorbonne Université, CNRS, ISIR
Abstract
Physical representations of data offer physical and spatial ways of looking at, navigating, and interacting with data. Whiledigital fabrication has facilitated the creation of objects with data-driven geometry, rendering data as a physically fabricatedobject is still a daunting leap for many physicalization designers. Rendering in the scope of this research refers to the back-and-forth process from digital design to digital fabrication and its specific challenges. We developed a corpus of example dataphysicalizations from research literature and physicalization practice. This survey then unpacks the “rendering” phase of theextended InfoVis pipeline in greater detail through these examples, with the aim of identifying ways that researchers, artists,and industry practitioners “render” physicalizations using digital design and fabrication tools.————————————————————————-
CCS Concepts • Human-centered computing → Visualization techniques;
1. Introduction
Long before the invention of writing, people have used physicalforms to record information [Ins16]. Physical data representations–also called physicalizations – display data through the geometricor physical properties of an artifact [JDI ∗ ∗
15] to helpusers understand, explore, and perceive data. Research has shownthat physicalizations can improve the efficiency of information re-trieval and memorability of data when compared to similar designsshown on flat screens [JDF13,SSB15]; they can also positively im-pact data perception and exploration [TJW ∗ ∗ ∗
14, AFS05]), medicine (e.g., [BHR ∗
17, HAD ∗ physicalrendering process .Physical rendering –or rendering – makes the visual presenta-tion perceivable by bringing it into existence in the physical world[JD13]. This transformation of data through rendering is not oftena simple, straightforward process. Limitations of the fabrication’stechnology (e.g. size, speed and colour limitations) impose somerestrictions in the transformation. Physical rendering requires aninterdisciplinary understanding of how data is represented and vi-sualized (Visualization and Computer Graphics), how to design andcreate physical objects (Design and Fabrication), and how peoplephysically interact with that data (Human-Computer Interaction).In this survey, we focus on the rendering phase of the extendedInfovis pipeline [JD13] and review approaches and methodologiesfor converting data into digitally-fabricated physicalizations. ThisSTAR aims at addressing the following questions: • What is the target dataset and the resulting visualization idiom ,i.e., the distinct approach to create and manipulate the visual rep-resentation [Mun14]? • What are the dominant strategies/approaches towards physicalrendering? • What are the challenges of rendering transformation?Our goal is to provide physicalization researchers and design-ers with a review of alternative physical rendering methods andtheir trade-offs, such that they can select rendering methods tai- © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and JohnWiley & Sons Ltd. Published by John Wiley & Sons Ltd. a r X i v : . [ c s . G R ] F e b javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process Fabrication in Architecture and Art:CUMING CADSMI FASEFor each candidate
Final corpus object made after 1990
Systematicsearch of academic work
Artists and practitioners and more academic work listed on dataphys.org
HCI:CHI ProceedingsVisualization:IEEE VISEuroVisIEEE TVCGGraphics:EuroGraphicsSIGGRAPHSIGGRAPH ArtCG&ASearch terms:"fabrication","3D printing","physical visualization", "physicalization","digital fabrication" aims to visualize any type of data has a physical object as resultExclusion no noyes
Entries from artists and practitioners on dataphys.org/listCurated bibliography from dataphys.org/wiki/Bibliography no Exclusionmade using CAD and CAM? no object is passivedocumen- tation available yes no yes yes designed explicitely to show data nono yesyesyes Figure 1: Decision graph for the curating process of our corpus.lored to their goals and expertise. Although there exist other surveypapers related to various fabrication approaches [HIH ∗
13, BFR17,LEM ∗
2. Methodology
In this section, we discuss how we assembled our corpus of physi-calization examples for analysis.
Many academic and art communities explore the physicalization ofdata. We built a corpus from two sources: (a) a systematic literaturesearch and (b) specific physicalization examples from dataphys.orgOur systematic literature search started by filtering papers, shortpapers, and posters published between 2010 and 2020 that met a keyword search (CAD, modelling, data design, data-enabled de-sign, data-driven design, CAM, fabrication, 3D printing, compu-tational manufacturing, digital fabrication, physical visualization,physicalization, data materialization, embodied interaction, instal-lation, physical, physical material, prototype, rapid prototyping,shape-changing, spatialization, tactile, tangible, tangible user inter-faces, wearable, actuation, personal data) in the following academiccommunities: • Computer Graphics (Eurographics, SIGGRAPH, SIGGRAPHAsia, IEEE CG&A) • Visualization (EuroVis, IEEE Vis, IEEE TVCG), • Human-Computer Interaction (CHI Proceedings) • Fabrication in Art and Architecture (SIGGRAPH Art, SMIFASE, CUMINCAD).Meanwhile, we wanted to also include examples from thebroader art and design community whose physicalizations may notappear in academic literature. Dataphys.org has actively collectedexamples of physicalizations from various disciplines since 2013.We excluded work from before 1990 as CAD/CAM technologieswere less common. Also, we only considered examples with properdocumentation (e.g. published papers and reports).In the end, we gathered 250 examples, for which we filtered outthe entries that did not have any available documents explainingtheir physicalization process. This study was also quite helpful inmaking us more familiar with different communities working onphysicalizations.Once we established this initial corpus of data physicalizationexamples from academic and practitioner communities, we contin-ued to filter based on (a) availability of quality documentation withadequate detail to address research questions and (b) the use of digi-tal design (CAD) or fabrication (CAM) software and tools. We thenlooked at whether the physicalization was a passive object, or repre-sented through an active physical platform. We excluded any activephysical platforms that did not have specific data physicalizationapplications designed for them. © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Target data
Digital Design
Digital Fabrication
Active platforms
Augmentations
Classifications
Data Design Rendering Physicalization
Figure 2: Physical Rendering Pipeline with digital fabrication, presenting the main sections of the paper and techniques to digitally fabricatea physicalization. Each square represents a main step, sections can be identified below their name in Italic.A summary of the paper collection and corpus curating processis presented in Figure 1. Our final sample includes 137 works – 75long papers, 17 short papers and posters, 4 thesis and dissertations,27 works presented on websites, 13 videos. 96 physicalizations aredesigned and developed by academic groups and researchers, 37projects are made by artists and practitioners, and the professionalcommunity, such as architects, were also part of the physicalizationcommunity by making 4 projects. Our corpus and its analysis areavailable to readers as static tables included in the paper (see Ta-ble 3 and Table 4), as well as an interactive online version underhttps://yvonnejansen.github.io/physicalization-rendering/.
The process of physicalization includes some actions and activi-ties such as collecting data from different types, digitizing data andconvert it to a visual form, fabrication, etc. We structure the maincategories of our coding schema into a process pipeline of physicalrendering during the fabrication of a physicalization in Figure 2.This pipeline represents some of the possible steps coded in thecollection to digitally fabricate a physicalization, from the data tothe final artefact. Each step represents a section of the paper.
3. Physicalization Classification Schemes
Our corpus contains a breadth of physicalizations that haveemerged from different communities (research, art, design) withdiverse skill sets, intentions, and approaches to physicalization.In this section, we discuss the breadth of our corpus along withseveral factors: information and scientific visualization; pragmaticand artistic; passive, active, and augmented physical objects. Wealso discuss application-centric and idiom-centric classifications ofthese physicalizations. Note though that categories within thesefactors and classification schema are not mutually-exclusive, andsome physicalizations can be described as simultaneously address-ing multiple categories.
Physicalizations can be categorized by a classic method of clas-sifying visualizations: distinguishing between Information Visual- ization (InfoVis) and Scientific Visualization (SciVis). This distinc-tion is, however, elusive, difficult to define, and controversial withinthe visualization community. One definition of the distinction be-tween InfoVis and SciVis by Tamara Munzer: “it’s InfoVis whenthe spatial representation is chosen, and it’s SciVis when the spatialrepresentation is given” [Mun08].Following this definition, our corpus includes 27 papers andprojects that can clearly be categorized as Infovis and 33 thatcan clearly be categorized as SciVis (see Table 3 and Table 4).Both categories tend to not focus on specific types of data andinclude a wide variety of examples. InfoVis physicalizations inour corpus include the representation of country indicators to ex-plore correlations between data series [Dwy04], personal activ-ity data [STS ∗ ∗ ∗ ∗ We also looked at whether a physicalization was created in pursuitof pragmatic or artistic goals. We adopted Robert Kosara’s inter-pretation of pragmatic visualizations as having “the goal [...] toexplore, analyze, or present information in a way that allows theuser to thoroughly understand the data” and of artistic visualiza-tions as having the goal “to communicate a concern, rather than toshow data” [Kos07]. In our classification, we considered physical-ization examples representing data in a playful manner, to expressconcerns, or to offer inspiration as artistic, regardless of whether ornot they were made by artists.However, the distinction between pragmatic and artistic phys-icalizations is blurry. Examples like a piece from the piechart [Rüs14], a robotic pie-charts-on-pies machine, uses a clas-sical encoding (pie charts) in an art exhibition with the intent todraw attention to gender distributions in the tech world. This exam- © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process ple is simultaneously pragmatic (allowing the viewer to thoroughlyunderstand the data) and artistic (made with the intent to commu-nicate a concern). Our corpus includes around 10 artistic physical-izations and 85 data sculptures listed in the dataphys.org/list siteas well as the SIGGRAPH Art track. Many of these examples arepragmatic as well.
Another dimension on which physicalizations can be classified ishow they employ computational components. Many physicaliza-tions are disconnected from all types of computational machinesonce fabricated. We call these physicalizations passive in line withprevious work [Jan14]. Note that passive only refers to the use ofcomputational power and not to the support of interactivity moregenerally. We discuss in Section 6.4 how different fabrication andassembly techniques can permit different levels of (manual) inter-actions such as sorting and filtering [JD13]. Our corpus includes 79examples of passive physicalizations.In 13 examples, we observed the combination of passive phys-icalizations with augmentations such as projections or augmentedreality overlays which provide access to computational function-ality on some of the data dimensions. For example, Gillet andcolleagues [GSSO05] presented physical molecule models whereusers can explore the interaction of their electromagnetic fields inaugmented reality when the molecules are brought close together.In another example, Hemment and colleagues [Hem13] augmentedphysical height maps of Twitter sentiments about the 2012 OlympicGames by projecting on top of them and thus enabling visitors tohighlight different aspects of the data interactively. We discuss aug-mented physicalizations in more detail in Section 5.4.Finally, we identified 47 examples of physicalizations that aredependent on some form of computational or at least electricalpower to show their data to an observer. There are many differ-ent ways of realizing this which we review in Section 5.5 in moredetail. Using active rendering techniques not only enables the ad-dition of some computer-supported interactions – as with aug-mented physicalizations – but also supports functionalities such asupdating or loading different data sets (e.g., [HGG ∗
16, THK ∗ ∗
18, FLO ∗ One possible method of classifying physicalizations is through dif-ferent applications that they can be used for. For instance, somephysicaliztaions are designed to simplify the understanding ofinformation or scientific data and help a specific group of prac-titioners or general public easier understand such concepts. Suchphysicalizations raise awareness, help in making better decisions,and can be used as collaboration tools among various professionalor academic groups (e.g., [TL16, PGDG12, ASS ∗
19, KHT ∗ ∗ raise selfawareness (16 works in our corpus). Many of such physicaliza-tions focus on personal activity and health tracking data that wewill discuss in Section 4.3. Another goal for making such phys-icalizations has been keeping track of progress during PhD stud-ies [KS12, SSJ ∗ improving accessibil-ity , such as tools for helping people with limited or no vision (e.g.,[PTPM17, TGZ18, SRK ∗ learning and education (e.g., [DMAS17, BKW ∗ research and engineering tools (e.g.,[MIWI16, ŠLH ∗
14] (with 12 total examples), and for presurgicalplanning (e.g., [BKW ∗ Munzner calls every distinct approach to create and manipulate avisual representation from the abstract data an idiom [Mun14]. Sheintroduces two major categories in idiom design: visual encodingidiom, i.e., representational idiom, and interaction idiom. The vi-sual encoding idiom controls what people see in a visualization.Based on the physicalizations reviewed in our corpus, a high-level categorization of representational idioms can be introducedas follows: physical charts, topography and elevation models, in-formative spaces and installations, and unique data objects.
Physical Charts.
Munzner’s visual encoding idioms reflect differ-ent graphical chart types (e.g., bar charts, line graphs, etc.). Manyphysicalizations extend visual encoding idioms from graphical rep-resentation into physical 3D objects. These include physical barcharts (e.g., [SSJ ∗ ∗ Topography and Physical Elevation Models.
Physical ElevationModels generally physicalize elevation data, terrains and topogra-phies (e.g., [TMH ∗ ∗ ∗ Informative Spaces and Installations.
These physicalizations aremostly architectural spaces or artistic installations, designed withdata, for the purpose of conveying a message. Architects and de-signers now use computational design methods to leverage avail-able data streams and generate novel forms and spatial opportu-nities [BM17, GR17]. Physicalizations with this representationalidiom aim to provide an atmospheric experience for users while © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 3: Using Physical Elevation Models for showing the averageprices for building lots in Germany (Left) and time-distance to thenext airport (Right). Images taken from [Ras11].reflecting a message from their target data. In such work, com-municating information and producing abstract effects (e.g., withlights, colours, movements) are mixed in the form of an installa-tion at an architectural scale (i.e., to form a space). Such approachhelps in mixing the didactic and literal representations with qualita-tive and atmospheric experiences. Didactic spaces are also referredto as data spatialization [Mar14]. For instance,
Data-spatializedPavilion [HDA ∗
19] introduces a novel method to make a data-driven pavilion through catoptric (mirror-assisted) anamorphosis,where the input data defines the physicality of the pavilion and si-multaneously remains readable. In another example,
Weather Re-port [KJA ∗
18] uses a set of two illuminated balloon walls, one forrepresenting real-time weather data (quantitative) and one for visu-alizing the audience’s memories of weather (qualitative). There are8 examples of informative spaces in our corpus and 15 examples inthe form of active installations.
Unique Data Objects.
Unique objects designed with data – fre-quently referred to as data sculptures– can take many forms, shapes,and scales. Many of the physicalizations in our corpus are ob-jects small enough to be picked up and held. For example,
MotusForma shows 10 hours of movement trajectories in the lobby areaof Pier 9 [AS16]; Doug McCune’s physical maps show data relat-ing to living conditions in San Francisco [McC13, McC16]; LorenMadsen’s data sculptures represent the increase of cost of livingfrom 1960 to 1994 [Mad95]. Some physicalizations were createdas wearable clothing [Per14, CO14] or jewelry [Kan17, LCN15].Some artists have taken unique approaches to make data physical.For instance, the
Snow Water Equivalent Cabinet shows snowpackmeasurements of the years 1980-2010 by making a drawer-like ply-wood sculpture, where the size of each drawer corresponds to theannual precipitation by year [Seg11].
4. Target Data for Physicalization
Many types of datasets have been transformed into physicaliza-tions, from personal activity data [KHM14,STS ∗ Due to the complexity and delicacy of medical and biologicaldatasets, tangible visualizations that can show different modes ofsuch datasets for a range of stakeholders can be quite useful. As aresult, physicalization for these datasets has been broadly studiedand practiced.In [GWW ∗
04, GSSO05], Gillett et al. combine 3D printing andvirtual reality to improve learning complex biological moleculestructures; using their system, people manipulate a physical 3Dprinted model that is tracked by a camera, controlling the view-point of a graphical visualization displayed on a screen. Rezaeianand Donovan represented the personal DNA data of individualsas 3D printed jewelry [RD14]. Variety of datasets including MRIhas been 3D printed in plausible forms using multimaterial voxel-printing method in various colors (see Figure 5a) [BKW ∗ ∗
14] turns individual’s bio-data (e.g., Galvanic SkinResponse (GSR) and Heart Rate (HR)) into a colorful 2D paint-ing. Personal health data is physicalized in [FF14] through a multi-modal representation. For instance, a two dimensional wooden ra-dial display that simultaneously visualizes temporal heart rates andskin temperature (see Figure 5b). Nadeau and Bailey created 3Dphysical models with interlocking pieces from medical volumet-ric data via solid free-form fabrication equipment [NB00]. Thrunand Lerch used 3D printing to represent high-dimensional datasetssuch as pain phenotypes as a landscape in four different colors(i.e., white, red, green, blue, yellow), highlighting distance [TL16].Ang et al. [ASS ∗
19] physicalized blood-flow datasets by 3D print-ing slices of curves or glyph to resemble flow directions in a vol-ume (see Figure 5c). Lozano-hemmer physicalized viewers’ heartrates with a set of light bulbs hanging in a room, synchronizingthe bulbs with each heart rate as viewers began interacting with thework [LH06].Geurts and Guglielmetti [GG15, Geu18] discussed the possi-bility of capturing thoughts and the relationship of cognitive andemotional to one’s work and living environments in digital andvisual forms (e.g., images). Neural connections in the brain aresimulated and physicalized by a set of bottles spinning on a tableforming various patterns [LH04a]. To promote physical activities,EdiPulse [KAP ∗
17] transformed self-monitored physical activity © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 5: Examples of biological and medical data physicalizations. (a) Diffusion-weighted MRI data capturing the diffusion of watermolecules in white matter brain tissue. Image taken from [BKW ∗ ∗ Engaging physicalizations can be very helpful for communicatingstatistical datasets with the audience [JDF13, Mar14]. Statisticaldatasets are usually quantitative values represented in numericalor string formats. Examples of such datasets include water con-sumption (in million gallons per day) [Seg11], class sizes and thenumber of graduates [Mar14], etc. Here, we list specific examplesof statistical and mathematical datasets from our sample.Taher et al. created responsive bar charts to communicate statis-tical data (e.g., international export data) with rods and RGB LEDs[THK ∗
15, TJW ∗
16] (see Figure 6a,b). Pulse [FM12] is a tangibleline graph composed of a string whose position is modified by sixservo motors. Drip-By-Tweet [Str14] visualizes the statics relatedto a voting mechanism collected on Twitter by a series of tubeswhose amount of fluid changes based on the number of cast votes(see Figure 6c). In Tape Recorders [LH11], motorised measuringtapes visualize the amount of time that visitors spend in a partic-ular installation (see Figure 6d). Kauffman and Brenner [KB13]created a physicalization of high school drop outs in New York byhighlighting the locations of schools on the map with a set of beads.The beads are connected to a string below with lengths relative tothe number of students who dropped out.To raise awareness about the lack of female representation inart and tech,
A Piece of the Pie Chart transformed gender ratiosinto real, edible pie charts [Rüs14]. Floating charts [OPSR16] isan acoustic levitation display for placing free-floating objects thathas been constructed to visualize a dynamic floating chart to reflectchanges in data.Le Goc et al. [LGPF ∗
18] introduced Zooids, a dynamic phys-icalization where small moving robots form patterns and clustersrepresenting data points to facilitate decision making (e.g., rank-ing applicants for departmental admissions). Emoto [Hem13] usedorigami-like data sculptures to communicate Twitter data related to London 2012 Olympics events. Fantibles [KAL ∗
16] is a per-sonalized memorabilia capturing an individual’s commentary aboutsports (e.g., cricket) through a nested double-ring physicalization.Starrett et al. [SRP18] turned the famous computer graphics ob-ject, Utah teapot, into a visualization by changing its base to a curverepresenting datasets by intersecting circles. Chaotic Flow [LJL12]is an installation of colorful flowing liquid that visualizes the flowof Copenhagen bikes. Perovich et al. fabricated lace patterns forclothes based on air pollution datasets [Per14]. McCune createdphysical maps physical thematic maps to turn “horrible data” (e.g.,murders or natural disasters) into visually pleasing physicaliza-tions [McC13]. Cosmos [JG14] is a spherical wooden sculpturethat represents data from forests that describe the take-up and lossof carbon dioxide by trees. Data Moiré [HC17] is an effort to phys-icalize the data on IBM Digital Analytics Benchmark to a large-scale feature wall that is CNC-machined. Madsen also representedthe evolution in the world population from 10,000 BCE to today asa 20-meter long data sculpture [Mad95].Radically different materials and forms have been used for mathdataset physicalizations such as crystal engraving [Bou15] or pa-per [DeM11]. For instance, to facilitate students with visual im-pairment to learn math, VizTouch has been developed to produce3D printed tactile visualizations to represent mathematical con-tents such as graphs [BH12]. Wavefunction [LH07] uses a set ofchairs (50-100) that are arranged like a regular array of rows. Theheight of these chairs change when an audience approaches a chairproducing a crest and the height change propagates through otherchairs.
Self-monitoring practices raise awareness about an individual’spersonal habits; as a creative representational method, physicaliza-tions can encourage different groups of people to actively monitortheir progress and become conscious about their habits and behav-iors, such as physical activity [KHM14]. Towards this goal, Stusaket al. designed a system that collects datasets from users’ runningactivity (e.g., duration, distance, elevation gain, average speed) andgenerates multiple types of activity sculptures [STS ∗ Patina En-graver uses the gradual development of patinas to map user activitydata to a wearable wrist band by applying stippling technique (i.e., © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 6: Engaging physicalizations help significantly in conveying the message of mathematical and statistical datasets: (a) Actuating phys-ical bar chart with LEDs to show international export data. Image taken from [THK ∗ ∗ TastyBeats prepared drinks for users af-ter a workout, based on their heartrate values [KLA ∗ ∗ Go and Grow mo-tivated tracking and self-reflecting on their fitness data by mappingactivity data proportionally to the amount of water given to a liv-ing plant; the more active the plant owner, the healthier their plantsbecome [BPAC16].With 13 works in this category, physicalizations that reflect per-sonal data show an emerging and interesting direction for fur-ther exploration. Moreover, the studies on personal physicalizationsdemonstrate how engaging idioms (food, plant growth, wearableobjects) can encourage and motivate physical activity and providepleasurable interactions with personal data. As a deeper investi-gation of the intersection of personal data and materiality, Khotet al. [KHM20] reviewed examples of personal physicalizations topropose a conceptual design framework for creating material rep-resentations of physical activity data.
Geospatial datasets are well suited for fabrication as they refer to aparticular spatial location or geographical scene. Therefore, manyworks benefitted from different physicalization approaches to bet-ter represent such datasets.Geospatial datasets are typically of four main formats: imagerydatasets (e.g., satellite images), elevation datasets (e.g., DEM), vec-tor datasets (e.g., roads, boundaries), or 3D geometries (e.g., 3Dbuildings) [MAAS15]. Various forms of geospatial physicaliza-tions have been developed for the purposes of education [KB14,MDES21], providing scenery models or data [Ras11], or raising awareness [Kil14]. In the following, we discuss such approachesand provide details about their methodology.Tangible Landscape is a 3D educational physicalization to teachtopography (i.e., the shape of terrains) [MTP ∗ ∗
10] integrates a laser scanner, projector, and a flexiblephysical 3D model; end-users can control a digitally projected sim-ulation by add and remove artifacts on the 3D model. Created formilitary purposes, Xenotran [Sch04] is a self-reconfigurable solidterrain model whose surface movements are controlled by 7000 ac-tuators.Geospatial physicalizations have also been used to address inter-esting applications: depicting a case study of a plane crash [Inc03],showing parks and forests in Berlin [Mei17], visualizing worldpopulation density [Bad13], and showing people movements ina lobby space [AS16]. In addition, we found examples of artis-tic geospatial physicalizations, such as the data-spatialized pavil-lion [HDA ∗ Environmental data addresses measurements of the environment,its systems, and impacts on its ecosystem. Engaging visualizationsof environmental datasets is crucial to raise awareness about crit-ical issues including wildfire, global warming, animal extinctions,etc. Many of these examples are produced with artistic goals tooffer a critical perspective. Segal transformed the amount of wa-ter stored as snow throughout a season into furniture, where thechoice of forms and materials connected back to the origins of thedata [Seg11]. Aweida [Awe13] combined robotics and art to build © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 7: Tangible Landscape is a malleable model, equipped withprojectors, that enables users learn about various aspects of topo-graphical properties. Image taken from [MTP ∗ ∗
16] was aseries of physical ambient visualizations that let end-users to ex-plore and engage with environmental data. Data from Physikit was]visualized through movement (PhysiMove), vibrations (Physi-Buzz), air (PhysiAir), and light cubes(PhysiLight). Physicalizationdesigners have also leveraged the biological properties of plantsthat respond to environmental conditions to create human-readablestructures [YOC ∗ ∗ ∗ We also found examples of people transforming images and videointo physical artifacts in order to add tangibility or artistic fea-tures. Zhao et al. [ZLW ∗
16] produce artistic lampshades projectinggrayscale images onto surrounding walls. String Art replicates animage by several straight lines of strings that are tied to a set ofpins located on a frame [BRWM18]. Portal [HAA ∗
20] is a struc-ture produced by a laser cutter and a set of mirrors to create animage that does not exist in the environment by reflecting colorsfrom another given image. In addition, to produce paintings, water-color woodblocks are designed to ease the process of producingseveral copies of a painting [PPW18]. MoSculp [ZDX ∗
18] pro-duces a sculpture representing a moving object or person (e.g., adancer). Motion Structures [RG13] turns video frames (e.g., Gameof Thrones teaser) into 3D printed sculptures.
In 48 of our reviewed papers and projects, we found datasets thatdid not fit in the aforementioned categories. An example of these other datasets is motion, action, and movement, which can result ininteresting physical patterns rich in details. Motus Forma [AS16]captures 10 hours of people’s movement in a lobby space, with more than 1300 motion paths. By attaching sensors to the back ofcrochet hooks and combining the data into 3D coordinates via aProcessing script, Nissen and Bowers designed path-like patternsto capture hand movements of crochet practitioners with variedskill levels [NB15]. With the goal of understanding various activi-ties within a FabLab environment,
Cairn [GD17] is a collaborativesculpture with various laser cut pieces. Table 1 summarizes differ-ent types of datasets under the other category in this survey, alongwith their corresponding works.
Entry Dataset [AS16, NB15, GD17,LH04b, KGM ∗ ∗
16, ZYZZ15,TSW ∗
19, LMAH ∗ ∗
13, TVR ∗
12] 3D Patterns and 3D Objects[PTPM17, MIWI16,TGZ18, DLL ∗
15] Texture and Material[Rod18, KAL ∗ ∗
14] PhD Studies[Mar14] Different Degree Type Offered[MCG ∗
15] Astronomy[Som14, Epl12] Public Opinion[Hei15, Kou18, Gon16,GHHS14, Kat18] Words, Terms, and Text[Kis09, MP09] Emotions and Relationship Status[LGPF ∗
18] Tourist Peak Periods[TACS16, Kel09] Website Traffic[LGKP ∗
16, SRK ∗
16] Freehand Drawing[HKH ∗
04] Amount of Trash and Recyclables[BL12] FM Radio Spectrum[GYS ∗
12] Package Openability[ZC18] Taste StructuresTable 1: Other datasets used for physicalizations.
5. Design and Physical Rendering Approaches
In this section, we discuss methods used to make a visual presen-tation and bring it into the physical world. Our goal is to discussvarious approaches used for design and physical rendering, usingdifferent digital design and fabrication tools. Based on the reviewedworks in our corpus, a typical process planning for the physical ren-dering process consists of design sketching, making accurate 3Drepresentations of the physicalization design, AKA 3D modelling,physical prototyping, modifying the design (i.e., iterative design),final fabrication, and conducting studies (see Section 6 for iterativedesign and user studies). © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
In this report, physicalization design is the stage of making the ab-stract visual form and the final visual presentation, i.e., visual map-ping and presentation mapping as introduced in [JD13]. While thisstep is full of opportunities, it also introduces several challenges forvisualization designers who have always considered cognition andperception for their on-screen or paper-based designs. When work-ing in physicalizations, visualization designers should consider per-ception and experience of physical environments, materiality, cul-tural symbolism, and spatial relationships. Many of these chal-lenges have been explored and practiced for many years in the fieldsof industrial design and architecture. As a result, investigating thedesign principles and steps architects and industrial designers takecan be quite helpful for the design of physicalizations as well. Sosaet al. have introduced four design principles inherited from indus-trial design that can be applied to physicalizations [SGE ∗ Design is the first stage of the rendering process that applies visualmapping transformation to data and gives it an initial visual form.The introduction of CAD and its ability to deal with more complexgeometrical problems [Kha10] has made digital design a popularapproach for physicalization. CAD, as an umbrella term, covers avast array of tools that produce different results such as 2D draw-ings and 3D models. CAD data has the great option of transfer-ability into other software platforms to control the appearance andother formal characteristics of physicalizations [Dun12].
For various physicalization scenarios, a 2D drawing needs to bemade in CAD. This 2D drawing can be either a continuous path(vector) or a discrete path (raster), such as a series of images. It isthe output of the processing pipeline of the fabrication techniquethat should be considered for making decisions about creating vec-tor or raster designs (e.g., cutting lines vs. engraving images in lasercutting).2D CAD is usually used for preparing outlines and contour linesto be used for laser cutting, such as the pieces making the
Trend inWater Use sculpture [Seg11] or tokens representing people’s activ-ities in FabLabs in
Cairn [GD17]. One of the frequently used CADsoftware to make vector 2D drawings for physicalization purposes is Adobe Illustrator. For instance, Häkkilä and Virtanen have trans-lated the collected sleep data from an Oura ring to 2D charts and2D paths for laser cutting, using Illustrator [HV16]. There may besome design cases for physicalizations that hand-drawn sketches ofpaths should be translated into vector data. In such cases, Illustra-tor can be used to trace over scanned hand-drawn paths, such asthe 1306 individual paths showing the movement of people in Mo-tus Forma [AS16]. Outputs from programming-based CAD designs(see Section 5.2.2) can be exported to Illustrator to make laser-cutready vector files. Such files include various line types, based on thedefined paths (e.g., cutting once or twice) and actions (e.g., cuttingor engraving) for laser cutters (see Section 5.3.3 for more details).An example of such application for Illustrator is
Blip , which hastransformed a year of travel into data sculptures [Gü11].Vector paths created by 2D CAD software can also be used aspart of the modelling process in any 3D CAD platform to makevolumetric designs and generate suitable files for fabrication. Inthe following section, we will cover various scenarios for 3D CADmodelling that can be used for the design of physicalizations.
To model 3D objects that can be fabricated, three primary represen-tations are usually used: polygonal meshes, Non-Uniform RationalB-Splines (NURBS), and constructive solid geometry (CSG).Polygonal meshes provide a discrete representation in which anobject is represented by a set of polygonal facets indicating the con-nectivity of the shape along with a set of vertices with ( x , y , z ) co-ordinates providing the geometry. Due to the simplicity and effec-tiveness of this representation, meshes are industry standards andare included in many 3D modeling software programs includingMaya [Fou21] and Blender [Aut21b] and they have been also usedfor the sake of physicalization (e.g., [Bar11, Bar12]).To offer designers a higher degree of control on the form, dig-ital modelling programs also utilize continuous curve and surfacerepresentations in which a model can be modified by a set of con-trol points. NURBS are powerful representations in this setting ascontrol points can attain different weights to push or pull a curveor surface; a property that other representations such as B-Splinesdo not have and therefore they are limited in producing many sim-ple and complicated shapes including a circle. NURBS can be di-rectly used to create curves and surface patches. It is also possibleto make a 3D shape by attaching several NURBS patches or gen-erate a 3D surface from a profile curve using techniques such asthe surface of revolution or sweep surfaces (see Figure 9). Due tothese powerful features, NURBS is very popular in physicaliza-tion [TACS16, Kat18, HV16].Although the curves and surfaces produced by NURBS providea high degree of flexibility via control points and weights [Dun12],some designers, especially for designing CAD shapes, prefer to useCSG since it provides sharp and accurate final results. In CSG, ashape is produced by applying several operations (e.g., union, in-tersection, difference, etc) on simple shapes such as spheres andcylinders to produce an accurate final object. CSG has been alsoused for physicalizations such as the customized Lego-Bricks pro-vided by Schneider [KS12]. © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process Figure 8: (a) An example of a parametric design generated by Grasshopper. The visual programming interface of Grasshopper, as well asits data list, is shown on the left. Image taken from [Son21], (b) A diagram showing parametric design stages of generating form from theoriginal data, using Grasshopper. Image taken from [Mar14]Figure 9: Examples of NURBS surfaces generated by attachingseveral NURBS patches: (a) Surface of revolution, (b) Sweep sur-face. Images taken from [Sam13].In some physicalization scenarios, the 3D models are designed,developed, and made ready for fabrication by only using variousCAD software packages and their features and functionality. Ex-amples of such software programs are Maya [Fou21], 3D StudioMax [Aut21a], Blender [Aut21b], and Rhino [Ass21]. For instance,NURBS provided in Rhinoceros®, AKA Rhino, has been used tophysicalize various models (e.g., [TACS16, Kat18, HV16]). We re-fer to such modelling as
CAD in our taxonomy (see Table 3 andTable 4).To ease the process of modelling, many software programs pro-vide a Visual Programming interface, where users connect a se-ries of functional blocks into a sequence of actions. The onlyrequired “syntax” in such method is that each block should re-ceive the appropriate data types as its input. Such solution is re-ferred to as parametric design [Dun12]. Note that this term isdifferent from parametric representation, such as NURBS and B-Splines, in which shapes are defined by benefiting from a param-eter space. As a rigorous rule-based system, parametric design in-volves precise, step-by-step techniques that make multiple optionsbased on a set of rules, inputs, and values specified by design-ers [Dun12, Jab13]. Grasshopper®, a visual programming plug-in designed for Rhino®, is one of those mediums that has a vi-sual interface and its components can provide, manipulate, and modify data, as well as draw and modify objects (see Figure 8).Grasshopper has been extensively used to produce physicalizationtechniques [Awe13,VTOS14,HDA ∗ Parametric Design in our taxonomy (see Table 3and Table 4).User interfaces for 3D modeling commonly follow the WIMP(Windows, Icons, Menus, Pointer) paradigm [JS11]. Sketch-basedinterface is considered as an alternative paradigm for 3D modeling[OSSJ08]. In this approach, 2D hand-drawn sketches are used in themodeling process, from model creation to editing and augmentingthe initial model in an iterative manner [OS10, OSSJ05].Extra development and customization sometimes have been em-ployed as pre-processing, post-processing or in the form of script-ing to prepare data or add necessary functionality. For example,Processing [FR20] has been used to produce line graphs of voterapproval rate data, available on the Internet, before making 3Dshapes for fabrication [Epl12]. To physicalize geospatial datasets,the coarse geometry of the Earth has been first extracted from aDigital Earth platform and then Rhino is used to develop the forms,design data attachment details, and make the pieces fabrication-ready [DMAS17,MDES21]. Scripting has been performed to make3D models and hinges for producing a mathematical puzzle ben-efiting from CSG operations available in Blender [LMAH ∗ ∗
20] have used Grasshopper and cus-tom Python scripting to build Portal. In our taxonomy table (seeTable 3 and Table 4), we have referred to such design approach as
Hybrid .There are many cases in the design of physicalizations whereoff-the-shelf CAD software, and even parametric or hybrid de-sign approaches, are not able to handle the complexity of theprocess of transforming data into a model. In such cases, physi-calization designers make their own programs via available pro-gramming languages and libraries (e.g., C++ and OpenGL). Manydifferent programming languages have been used for physicaliza- © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process tion, among which Processing, an open-source Java-based lan-guage developed for designers, is the most popular. The Process-ing community has written more than a hundred libraries to fa-cilitate computer vision, data visualization, 3D file exporting, andprogramming electronics [FR20]. Depending on the community,other programming languages such as Python, Java, or C++ havebeen also utilized to make a customized modelling program. Phys-icalizations for which a standalone program has been producedinclude Landscaper [ADMAS18], works to add textures on 3Dprints [SPG ∗
16, ZYZZ15, MAWS15], make water color paintings[PPW18], etc.
Fabrication makes the visual presentation perceivable by bringingit into existence in the physical world [JD13]. In digital fabrica-tion, computer-controlled manufacturing machines receive digitalmodels to build 2D or 3D objects [SSJ ∗ ∗ ∗
14, HIH ∗ A trivial technique to bring patterns, designs, and visualizationsinto the physical world is traditional (2D) printing. In 2D printing,key parameters are the print resolution and the printer gamut de-fined by the inks or toners employed [HIH ∗ The general concept in additive manufacturing is to build objectslayer-by-layer from a small number of basis materials [HIH ∗ Technique Attributes
Cutting • Easily accessible, • Makes shaped 2D elements from sheet mate-rials • Cutting Methods: Laser, Water Jet, PlasmaArcSubtractive • Takes material from an existing solid volumeand creates the desired shape, • Axially, surface, or volume-constrained cut-ting heads • Advantages:1. Larger component size,2. Wider range of material selection,3. More precise fabrication,Additive • Converts CAD to a series of 2D layers, i.e.,layer-by-layer fabrication (AKA rapid proto-typing) • Advantages:1. Direct “file to fabrication” process,2. Fabricates complex forms,3. Non-expert use, • Disadvantages: limited size, limited range ofmaterials, lengthy production times • Examples include: 3D printing techniques(Fused Deposition Modelling (FDM), Stere-olithography (SLA), Direct Metal Laser Sin-tering (DMLS), Selective Laser Sintering(SLS), Selective Laser Melting (SLM), Elec-tron Beam Melting (EBM)), knitting ma-chinesFormative • Uses mechanical force, heat, and steam to re-shape • Can be axially or surface constrained • Examples include: vacuum forming, thermo-forming (after 3D printing)Table 2: An overview of digital fabrication tools and techniques.One of the most common tools that digitally fabricate objects withan additive approach are 3D printers. Over the past decade, 3Dprinters have become more accessible to the consumer market withlow maintenance and operating costs. Moreover, the possibility ofmaking complex objects by using 3D printers have made them acommon choice for making prototypes or final physicalizations.Different types of 3D printers exist that all build objects on a layer-by-layer basis, but some locally deposit material and some solid- © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 10: FDM 3D printing has been used to produce physicalizations of health, personal, and sports data: (a) The Hypertension SingingBowl is a stainless steel 3D printed sonification that has transformed blood pressure data to a sculpture that rings. Image taken from [Bar14],(b) A figure and a necklace sculpture physicalizing running activity. Image taken from [STS ∗ ∗ ∗ ify material within a non-solid substance [LEM ∗ FDM 3D Printing.
One of the most accessible and affordable3D printers are Fuse Deposition Modelling (FDM) printers thatmake 3D objects layer-by-layer through heating and extrudingthermoplastic or wax filaments [ZDS16]. FDM 3D printing has along history in physicalizing complicated shapes such as macro-molecular assembly [BSJ98]. Until now, many physicalizationshave been produced via FDM printing for different applicationssuch as education [BH12, MDES21, KB14], project management[KS12], producing geological artifacts [HDA ∗
19, McC13, AD-MAS18, LBRM12, DEBS18, KHT ∗ ∗ ∗ ∗
18] (see Figure 11), environmen-tal data [Whi09], astrophysical [MCG ∗
15] and statistical data[McC16, GYS ∗ ∗ cm . There-fore, breaking a large model into printable volumes have been em-ployed [ADMAS18,LBRM12]. In addition, the results of FDM areusually limited in terms of number of colors, therefore innovativesolutions have been proposed to overcome these challenges. To re-solve this problem, geological features with different properties are Figure 12: One solution to overcome the lack of colour in 3D print-ing is fabricating the physicalization in discrete pieces, each witha different filament colour, and assemble them. Image (a) takenfrom [ADMAS18] and image (b) taken from [TL16].printed in different but limited colors [TL16, ADMAS18] (see Fig-ure 12). Projectors have been also used to visualize data on a basemodel [DEBS18]. Layer Solidification 3D Printing.
Layer solidification is a 3Dprinting process in which the top (or bottom) surface of the objectis solidified from a non-solid material, such as liquid or powder,within a tank. This process is executed by vat photopolymerization(e.g., stereolithography or SLA 3D printers), powder bed fusion(e.g., Selective Laser Sintering or SLS 3D printers, binder jetting(e.g., plaster powder binding), and sheet lamination (e.g., paper lay-ering–cutting) [LEM ∗ ∗
16] and data sculptures [Som14,SRP18], reservoir fieldexploration [NLC ∗ ∗
18] and sound[Bar12], even cooking molds [ZC18] (see Figure 13).
3D Colour Printers.
Since colors play an important role in anunderstandable visualization, color 3D printers (e.g., ZCorpora-tion multi-colour 3D printer) have been used to produce geologi- © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 13: SLA 3D printing has been used to make plastic molds(a) for silicone casting (b) to bake cakes representing various tastestructures (c). Images taken from [ZC18].Figure 14: Chronofab uses Stratys 6-colour 3D printers to fabricatemotion. Images taken from [KGM ∗ ∗
18] or motion[KGM ∗
16] (see Figure 14) and Connex3 500 has been used tophysicalize a variety of creative objects that can handle deforma-tion or attain specific textures [PTPM17].
Subtractive manufacturing techniques are on the opposite side of3D printing. In other words, rather than incrementally building upa model, subtractive techniques gradually remove material from anunmachined part by using a sharp cutting tool [LEM ∗ CNC.
Computer Numerical Control (CNC) is one of the mostcommonly applied methods of digital fabrication [AFS05, Kol04,Dun12], used by 10 projects in our corpus. CNC has the potential tofabricate double-curved and developable surfaces [Kol04, AFS05].In CNC milling, stepper motors control the movement of the in-dividual axes of tool movement. Two types of artifacts are com-mon with CNC machines: step artifacts and tool path artifacts thatleave tiny grooves on the final model [LEM ∗
17, MAYZ ∗ ∗ ∗ Cutting Techniques.
Cutting techniques can be considered as asub-category of subtractive methods. One of the most popular cut-ting methods is laser cutting (used by 20 works in our corpus),mostly due to its speed, efficiency, and its ability to cut a widerange of materials [SSJ ∗ ∗ Formative fabrication processes utilize mechanical forces to re-shape or deform materials into the required shape. Examples offormative approaches are vacuum moulding (i.e., heating a thermo-plastic sheet of material until it becomes malleable and then suck-ing it on a shape using vacuum pressure) and thermoforming (i.e.,heating a sheet of plastic material until it becomes malleable andthen forming the sheet onto a forming core shape). ComputationalThermoforming [SPG ∗
16] introduces a novel method for the fab-rication of textured 3D models. This approach is meant to be usedfor customized, unique objects, which makes it a useful solutionto support colour and texture for physical rendering of physicaliza-tions. Figure 16 illustrates the whole process of transferring a 3Dmodel into a plastic replica with the original texture applied atop it.
In our survey, we refer to the fabrication method of a work as hy-brid when a series of various methods have been used to make onesingle physicalization. In other words, if a physicalization systemproduces different results, each with one single fabrication tech-nique, it will not be counted as a hybrid method in our work.There are 13 examples of hybrid fabrication methods in our cor-pus. In some cases, hybrid approaches have been taken to deal with © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 15: Subtractive techniques, such as CNC and laser cutting have been used in many physicalizations: (a) Sculpture carved out of a blockof wood by a CNC milling machine, showing wind directions. Image taken from [Kna12], (b) Metal panel cut by a CNC machine to showIBM sales in the form of a data spatialization. Image taken from [HC17], (c) An example of shop drawings and technical assembly diagramsfor physicalizations that are fabricated in pieces and require assembly. Images taken from [Mar14], (d) Various woodblocks prepared by lasercutting to make a watercolour painting of a flower. Image taken from [PPW18], (e) Carved out parts of an MDF sheet are filled with moss tomake a living map of forests. Image taken from [Mei17].Figure 16: Computational Thermoforming is an advance methodto add colour and texture to physicalizations. Image taken from[SPG ∗ ∗ ∗ There are 74 works in our corpus that are fabricated in separatepieces and need to be assembled to form the final physicalization.Digital fabrication machines have a limited build area (for additivetools) and support specific sizes for sheets and blocks of material(for subtractive and cutting tools). This limitation forces the designto be either limited to a scale that can be fabricated in one pieceor to be piece-wise in a way that can be assembled and make abigger scale physicalization (e.g., [ADMAS18, Mar14, HC17]. As-sembling a physicalization also provides various interaction oppor-tunities (e.g., [GD17]) and can be used for educational purposes(e.g., [NB00, MDES21]) (see Section 6 for more details).
As introduced in Section 3.3, augmenting a physicalization adds anextra layer of information to an otherwise passive physical objectand can be a straightforward way of adding sophisticated interac-tivity without integrating actuators (e.g., motors) as is required foractive physicalizations. The augmentation approaches observed inour corpus can be summarized as a form of augmented reality, re-alized by projecting directly onto the physical object or through apersonal AR view using a head-mounted display or a hand-helddevice equipped with a camera. Overall relatively few of the itemsin our corpus (13 out of 137) fall into the category of augmentedphysicalizations.
Physicalizations using projection augmentation consist typically ofa passive, fabricated physicalization with additional data projecteddirectly onto it. In some cases, projections provide additional datalayers, such as annotations [PGDG12]. In other cases, they permitinteractivity, such as highlighting [Hem13]. Examples in our cor-pus include relief maps ( [TMH ∗
10, PGDG12, MTP ∗
18] (see Fig-ure 7), globe-based time-varying geospatial data [DEBS18] and adata sculpture ( [Hem13]) that shows Twitter sentiment data as anabstract relief heatmaps.When using projection augmentation, it is necessary to calibratethe physical object and the projection so that projected informa-tion lines up with corresponding physical features. The TanGeoMSsystem [TMH ∗
10] includes a combination of projection and 3D-scanning which enables the system to recognize the topology ofthe passive physical model and automatically detect how to ro-tate and scale the topological data to be projected onto it. For thePARM [PGDG12] and Emoto systems [Hem13], no calibration de-tails are provided; most likely, they require manual calibration be-tween the physical model and the projection. When projecting onnon-flat surfaces, it is often necessary to apply some form of pro-jection mapping [GI18] to avoid visible distortions of the projectedcontent. This is only discussed for the TanGeoMS system, wherethey found that a correction would only be necessary for heightdifferences of more than 6 cm, which did not occur in their case.The other two projection-augmented physicalizations did not dis- © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process cuss applying any remapping; this may be due to the small heightdifferences present in the two examples.
Unlike projected augmentation, personal augmented reality (AR)are technologies (head-mounted displays, mobile devices) that of-fer an augmented perspective of an object to a single individual.Augmentation through personal augmented reality is less presentin our corpus (only 3 items out of 137 examples make use of anindividualized augmented reality review). Of these, two are aca-demic works that date from before the consumer-availability ofaugmented reality headsets (2004/2005); these projects display theaugmentation layer on a separate display overlaid on a live videofeed of the physicalization [GWW ∗
04, GSSO05]. The third exam-ple, PLANWELL [NLC ∗ ∗ Active physicalization rendering techniques go beyond what is pos-sible through augmentation, but they are generally also more diffi-cult to realize. Nonetheless, they are more represented in our corpuswith 47 examples. Many of these examples fall into the category ofdata sculptures where artists explored unique ways to actuate ma-terials in some way suited to communicate their artistic intent orwhere academic authors sought to find unique ways of represent-ing often personal data in appealing ways. We focus here on exam-ples that are not entirely specific to the context in which they werecreated and whose analysis can inform the design of future activephysicalizations in some way.Overall, we identify three main goals for choosing an activerendering technique: supporting changes of a single dataset, sup-porting multiple datasets, and enabling interactivity beyond whatis possible using augmentation approaches. These goals are oftencombined although interactivity is less common for data sculp-tures/installations. Orthogonal to these goals is the question wherethe rendering technique should be capable of dealing with varyingnumbers of data points or whether those remain fixed once chosen.If an active physicalization only supports changes to a singledataset, then this suggests that the rendering technique was prob-ably specifically developed or tailored to that dataset and may onlybe applied to other data with difficulty. This is something we mostlyobserved with examples classified as data sculptures. For example, the artist Rafael Lozano-Hemmer developed multiple installationsfalling into this category using situated data like the presence andlocation of people in a room to actuate belts [LH04b] or tape mea-sures [LH11]. The latter example (shown in Figure 6d) uses actu-ated tape measures going up and down which resembles a bar chartand thus could also be used with different datasets. The number ofdata points of such a system remains fixed though and would needto be manually extended to be usable with different data sets. Bothof these examples react to people’s presence in the room, that is,sensors capture their presence and reflect the data on the shape andorientation of the system. Beyond that, these installations offer nointeractivity and they are purely meant to present data and not toenable onlookers to explore the shown data in any way.A few platforms have been proposed, mostly in academic re-search, which enable the visualization of various datasets as wellas interactivity aspiring to achieve a level of functionality knownfrom web-based visualization tools, such as, support to view differ-ent data, searching, filtering, highlighting etc. The development ofsuch platforms generally requires skills in mechanics, fabrication,sensor and actuator choice and placement, and micro-controller de-velopment. Reviewing all the issues related to developing new ac-tive platforms would go beyond the scope of this article. We reviewhere only active platforms included in our corpus.
Shape-changing displays are actuated devices capable of deformingin various ways [RPPH12]. One item in our corpus, the XenovisionIII system, is an actuated solid terrain model, commercially avail-able and marketed for military applications [Sch04]. It is capableof displaying any terrain data using its 7,000 actuated pins. Most ofthe other shape-changing displays in our corpus are created eitherby academics as proofs-of-concepts or by artists for installationsin museums or galleries. The most common form factor for suchdisplays are rods or bars arranged in arrays and capable of movingup and down to provide a 2.5D display [THK ∗
15, FLO ∗
13] (seeFigure 6a/b and Figure 17a). Such displays have generally muchfewer actuators than the Xenovision system, that is between 100and 1,000 pins. All of these 2.5D systems cannot display any over-hangs and only show data that could be represented by a 3D bar-chart resulting often in a resolution of 5 to 10 mm per actuator.The Relief prototype uses a similar principle but connected the in-dividual actuators with a cloth such that a smoothed surface is cre-ated [LLD ∗
11] which lends itself naturally to display terrain data.Shape-changing displays can also come in different base shapes.For example, Daniel and colleagues used a ring shape and actuatedtheir display such that rings could be stacked and each ring couldexpand its size to show different data [DRC18] (see Figure 17b).All of these displays can generally show different data sets or up-date the data being shown currently. Most also support interactivityin some way, often by covering the interactive area with a depthcamera and subsequently interpreting people’s gestures around thedevices.
While two-dimensional visualizations on-screen or three-dimensional visualizations in virtual reality are free to render data © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 17: Active physicalizations use different techniques to support physical changes: (a) inFORM is a shape-changing display that enablesseveral new interaction techniques. Images taken from [FLO ∗ ∗ A few examples in our corpus have used robotic arms to assemblephysicalizations. While the overall rendering platform can be con-sidered active – they take data as input and render a physical object– these resulting objects are, once assembled, passive objects. Ourcorpus includes three examples falling into this category. One usesa Kuka industrial robot to place nails in a substrate according towind data [Awe13], one uses a similar type of robot to span stringsto approximate the visual shape of the input data [BRWM18], and athird uses a self-built robotic system based on robotic toys and vac-uum cleaners to select already printed paper pie charts and placethem on actual pies [Rüs14].An entirely different approach was taken by Le Goc and col-leagues who developed a platform of small robots they calledzooids [LGKP ∗
16] (see Figure 17d). Each of these robots is meantto represent one data point and to move around a surface covered bystructured light within which it can orient itself and move to showdifferent facets of the data point it represents [LGPF ∗ The rendering process of a physicalization work, from the early de-sign stages to fabrication and assembly, is a skill-oriented approach.In other words, it demands knowledge and expertise in data visu-alization, digital design, and digital fabrication, and is sometimesinvolved with labour-intensive craftsmanship based on the appliedfabrication techniques. In order to overcome this issue, some re-search has been undertaken with the goal of automating the wholeor parts of the physical rendering process.
MakerVis [SSJ ∗
14] isone of the most inclusive platforms developed for this goal that iscapable of automating the whole physical rendering process fromdata filtering to physical fabrication. The prototype software ofMakerVis reads data in CSV or topoJSON (for prism maps) formatsand can produce data types that are compatible with CNC machin-ing, 3D printing, and laser cutting. The software is a web applica-tion built on top of NodeJS, D3, JQuery, and ThreeJS frameworks.Figure 18 shows the interface of MakerVis, as well as some resultsmade by it.
6. Discussion
In this section, we discuss some of the decisions and challenges oftransforming a data physicalization concept into a fabricated phys-ical form. First, we briefly discuss general digital fabrication issuesthat remain a challenge when rendering data physicalizations: de-sign for manufacture and assembly, and prototyping and iterativedesign. Then, we discuss challenges that impact both fabricationand data representation. When consulting with stakeholders, dataphysicalization designers are trying to understand both the physicaland representational requirements of the final object. When decid-ing on scale, the designer must balance scale limitations in manu-facturing against the readability or user experience of the physical-ization. Finally, assembly decisions can also limit users’ ability tointeract with data in the final object. © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Figure 18: MakerVis is an automated software developed for making physicalizations. (a) A screen-capture of the UI, (b) Physicalizationsmade by MakerVis. Images taken from [SSJ ∗ Many of the physicalizations from our sample struggled with chal-lenges that are universal to all digitally fabricated forms.
When optimizing a design for fabrication, designers often adjustthe features or complexity of a digital design to reduce manu-facturing time (e.g., [KHM14]). Many of the physicalizations inour sample required cutting and stacking layers of plastic or wood(e.g., [Hem13, Hus14, Stu13a, Seg11]). However, depending on thegeometric complexity of each layer and the number of layers, thisapproach can be time-consuming for both fabrication and assemblyof the final structure. We found two examples [DMAS17, Mar14]of physicalizations that included instructions to expedite the finalassembly.
To facilitate iterative design, physicalization designers used bothdigital simulation and lower-fidelity prototyping techniques to val-idate designs before following-through with the final version. Pro-totyping helps designers make better decisions about the specifi-cations of the physicalization; effective prototyping techniques re-quire the least amount of material or time investment to obtain in-sights to drive the next series of design changes.In a digital design process, digital simulations of the fabrica-tion process or the final object can facilitate faster iterations on aphysicalization design. This includes active physicalizations, whereKangaroo Plug-in for Rhino Grasshopper can be used to simulatemovements of active structures (e.g., [VTOS14,Awe13,TVR ∗ ∗
19, Bar11, Mar14], which offer quicker, lower-cost in-terim representations for design iteration. One approach is to builda small portion of the final physicalization, with the final scale,fabrication technique, and material as done by [ADMAS18]. An-other approach is to build a scale-model of a larger-scale physi-calization as a way of verifying the overall design and rationale(e.g., [HDA ∗ ∗ In the following section, we discuss three areas where digital fab-rication challenges and data representation challenges collide – un-derstanding users’ requirements; physical scale; assembly and in-teraction.
Some physicalization designers conducted formative user researchto guide what data needed to be represented by the physicaliza-tion, as well as any physical requirements of the physicalizationobject. This early decision-making process may include interviewsor consultations with stakeholders’ about their own understandingof their data or expectations on the form of a resulting physicaliza-tion [KHM14, VKBR ∗ ∗ ∗
18, LCN15]. For example, Gwilt et al.[GYS ∗
12] found that an engineering community preferred bar andpie charts, whereas a design community preferred data sculptures.Similarly, Khot et al. [KHM14] found their community of exer-cise enthusiasts preferred to represent their physical activity usinga non-scientific idiom (size of a frog) over a more scientific rep-resentation (physical bar chart). Meanwhile, molecular biologistsfrom Gillett et al. [GSSO05] preferred augmented physicalizationsover static physicalizations.
The physical scale of a physicalization fundamentally changes howpeople interact with it as an object; it also can introduce prag-matic constraints such as weight, balance, or portability of the fi-nal object. In room-size physicalization installations or spatializa-tions, such as [LH04b, LH11, BL12, Stu08]) and [Mar14, VTOS14,HDA ∗ ∗ ∗
10, ADMAS18], in-cluding augmented physicalizations [DEBS18,MTP ∗ ∗ ∗
19, BKW ∗ © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process provide very different interaction scenarios, as users can easily ma-nipulate them; objects that must be held in-hand (e.g., [PTPM17,TGZ18] can leverage tactile cues (surface roughness or texture).However, the scale of a physicalization also majorly influenceswhich fabrication techniques are appropriate. Most fabricationtechniques work at specific scales; for exmaple, most FDM 3Dprinters have an average build volume of around 20 cm . Larger-scale visualizations can be fabricated modular components at thescale of fabrication machinery, and then assembling these com-ponents into a larger structure. Decomposing a large 3D objectinto smaller parts that fit in the printing volume was introducedby Chopper [LBRM12] and then used by several physicalizationworks (e.g., [ADMAS18,DEBS18,DMAS17,FWF ∗ ∗ As with any digital design, when physicalizations are broken intoseveral pieces for fabrication purposes, the designer must definehow those pieces connect to each other. This includes definingwhich attachment techniques will be used, specific feature param-eters to assembly features (e.g., joint location, feature dimensions,part clearance between parts), as well as the assembly process it-self.However, the way in which a physicalization is designed to beassembled can determine what types of interactions end-users canhave with its represented data. For example, Jansen et al. [JDF13]created modular 3D bar charts that allowed the end-user to select,reorder, and independently compare datasets. If instead these barcharts had been glued together, the end-user would no longer beable to interact as deeply with the data itself. Similarly, the act ofassembling a dataset can require end-users to more deeply inter-act with the data itself. This is a guiding principle of both con-structive visualization [HCT ∗
14] and participatory visualization(e.g., [GD17]), but is also present in many other physicalizationsin our sample [KAL ∗
16, ADMAS18, KS12, DMAS17, MDES21].Landscaper [ADMAS18] requires assembling a series of geospa-tial features (e.g., trees, rocks, road networks, and urban features)which simultaneously requires the end-users to become familiarwith where each feature belongs within the space. In Nadeau etal [NB00], end-users could detach the 1:1 interlocking scale modelof the skull and brain segmentation to better understand the com-plex volumetric dataset via cutting operations; this cross-section isonly possible when this type of disassembly is pre-planned and al-lowed. Vol velle [SB16] introduced a novel interaction system thatphysically recreates the traditional concept of Volvelle.
7. Conclusion and Future Work
Physicalizations are effective tools for conveying the message ofvarious datasets and they can be rendered in many different meth-ods, for various applications, and in many representational idioms.Although many impressive works have been done in this field,many areas of rendering physicalizations remain unexplored. Forinstance, there have been many creative fabrication methods intro-duced by the computer graphics community with great potentialsfor physicalizations, especially for dealing with the issue of colourand texture. By applying such methods to physicalization render-ing, many interesting possibilities will be introduced. Also, num-bers in our corpus show that physicalization has been slightly un-derexposed in the scientific community and for pragmatic purposes.Many interesting aspects of physical rendering in artistic physical-izations and interactive installations have strong potentials to beapplied to scientific works as well. In our survey, we discussed thataugmented physicalizations add extra layers of information to pas-sive works and make them more sophisticated objects to interactwith. However, very few efforts have been made to develop suchworks. With the increasing popularity and technological advancesof devices that support augmented reality, physicalization design-ers should take advantage of elevating their passive designs to thenext level of informative and interactive representations.In this survey, we have provided an overview of physicalizations,their classifications, visual representation formats, and their targetdatasets. More importantly, we have reviewed various methods thatphysicalizations can be designed in digital design approaches andthen rendered physically by digital fabrication tools. We hope thatcomputer graphics, visualization, interaction, art, industrial design,and architecture communities find this survey useful and a sourceof inspiration to develop the physicalization field further. © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Basic Info Class Dataset(s) Category Digital Fabrication Digital Design P r ov e n a n ce Y ea r T itl e T yp e P a ss i v e A c ti v e A ug m e n t e d B i o / M e d i ca l M a t h P e r s on a l G e o s p a ti a l I m a g e / V i d e o S t a ti s ti ca l O t h e r I n f o V i s S c i V i s P r a g m a ti c A r ti s ti c D a t a S c u l p t u r e D P r i n t C N C L a s e r C u t R obo ti c A r m H yb r i d O t h e r A ss e m b l y ? C AD P a r a m e t r i c P r og r a mm i ng H yb r i d O t h e r I t e r a ti v e D e s i gn U s e r S t udy DataPhys.org 1995 [Mad95] Website • • • • N • N NDataPhys.org 1998 [BSJ98] Paper • • • • • Y • Y NDataPhys.org 2000 [NB00] Short P. • • • • Y • N NDataPhys.org 2003 [Inc03] Website • • • • N • N NDataPhys.org 2004 [LH04b] Video • • • • Y • N NDataPhys.org 2004 [LH04a] Video • • • • Y • N NSIGGRAPH 2004 [GWW ∗
04] Paper AR • • • N • N YDataPhys.org 2004 [Sch04] Website • • • • Y • N NDataPhys.org 2004 [HKH ∗
04] Short P. • • • • Y • N YDataPhys.org 2005 [GSSO05] Short P. • AR • • • • N • N YDataPhys.org 2005 [Dwy05] Thesis • • • • Y • N NDataPhys.org 2006 [LH06] Video • • • • Y • N NDataPhys.org 2007 [LH07] Video • • • • Y • N YDataPhys.org 2008 [Stu08] Website • • • • Y • N NDataPhys.org 2009 [Kna12] Website • • • • N • N NDataPhys.org 2009 [Whi09] Website • • • • N • N NDataPhys.org 2009 [LLD ∗
11] Paper • • • • Y • N NDataPhys.org 2009 [Kel09] Video • • • • • Y • N NDataPhys.org 2009 [Kis09] Website • • • • Y • N YTVCG 2010 [TMH ∗
10] Paper Projection • • • • N • N YDataPhys.org 2010 [GHK12] Website • • • • • Y • N NDataPhys.org 2010 [MP09] Paper • • • • • • • N • N YDataPhys.org 2011 [Bar11] Short P. • • • • N • Y NDataPhys.org 2011 [LH11] Video • • • • Y • N NDataPhys.org 2011 [Ras11] Paper • • • • N • N NDataPhys.org 2011 [Bow11] Website • • • • • Y • N NDataPhys.org 2011 [Gü11] Thesis • • • • Y • Y NDataPhys.org 2011 [DeM11] Website • • • • Y • N NDataPhys.org 2011 [Seg11] Website • • • • • N • N NDataPhys.org 2012 [BH12] Paper • • • • • N • Y YDataPhys.org 2012 [LJL12] Website • • • • Y • N NDataPhys.org 2012 [PGDG12] Paper Projection • • • • N • N YDataPhys.org 2012 [Hem13] Paper Projection • • • • Y • N NDataPhys.org 2012 [Epl12] Video • • • • • N • N NDataPhys.org 2012 [Bar12] Paper • • • • N • Y NDataPhys.org 2012 [FM12] Website • • • • Y • N NDataPhys.org 2012 [LH12] Video • • • • • Y • N NSIGGRAPH Art 2012 [BL12] Short P. • • • • Y • N YDataPhys.org 2012 [Len12] Website • • • Y • N NDataPhys.org 2012 [KS12] Short P. • • • • Y • Y YDataPhys.org 2012 [GYS ∗
12] Paper • • • • • Y • N YSIGGRAPH Art 2012 [Row12] Short P. • • • • Y • N NSIGGRAPH Art 2012 [TVR ∗
12] Paper • • • • Y • Y NDataPhys.org 2013 [Awe13] Video • • • • N • N NDataPhys.org 2013 [RG13] Website • • • • N • N NDataPhys.org 2013 [McC13] Website • • • • N • N NDataPhys.org 2013 [KB13] Website • • • • • N • N NCHI 2013 [JDF13] Paper • • • • Y • Y YDataPhys.org 2013 [Stu13b] Short P. • • • • Y • N NDataPhys.org 2013 [Bad13] Website • • • • • N • N NIEEE CG&A 2013 [TMB ∗
13] Paper • • • • N • N NSIGGRAPH Art 2014 [Rüs14] Short P. • • • • • Y • N NCHI 2014 [KHM14] Paper • • • • • N • N YDataPhys.org 2014 [KB14] Paper • • • • • N • N YDataPhys.org 2014 [Str14] Website • • • • Y • N NDataPhys.org 2014 [Som14] Video • • • • N • N NDataPhys.org 2014 [Hus14] Video • • • • Y • N NCHI 2014 [ŠLH ∗
14] Paper • • • • N • Y YSIGGRAPH Art 2014 [VTOS14] Paper • • • • Y • Y NDataPhys.org 2014 [RD14] Short P. • • • • N • N NDataPhys.org 2014 [LH14] Website • • • • Y • N NDataPhys.org 2014 [Per14] Thesis • • • • Y • N NDataPhys.org 2014 [Bar14] Short P. • • • • N • Y YDataPhys.org 2014 [CO14] Thesis • • • • • Y • N NDataPhys.org 2014 [GHHS14] Short P. • • • • Y • N NTVCG 2014 [STS ∗
14] Paper • • • • N • N YDataPhys.org 2014 [Kil14] Website • • • • N • N NCUMINCAD 2014 [Mar14] Paper • • • • • • Y • Y N
Table 3: Taxonomy of the works reviewed in this survey (Part 1). © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
Basic Info Class Dataset(s) Category Digital Fabrication Digital Design P r ov e n a n ce Y ea r T itl e T yp e P a ss i v e A c ti v e A ug m e n t e d B i o / M e d i ca l M a t h P e r s on a l G e o s p a ti a l I m a g e / V i d e o S t a ti s ti ca l O t h e r I n f o V i s S c i V i s P r a g m a ti c A r ti s ti c D a t a S c u l p t u r e D P r i n t C N C L a s e r C u t R obo ti c A r m H yb r i d O t h e r A ss e m b l y ? C AD P a r a m e t r i c P r og r a mm i ng H yb r i d O t h e r I t e r a ti v e D e s i gn U s e r S t udy DataPhys.org 2014 [JG14] Video • • • • N • N NCHI 2014 [SSJ ∗
14] Paper • • • • • • • Y • N YDataPhys.org 2014 [FF14] Short P. • • • • • N • Y NCHI 2015 [YOC ∗
15] Paper • • • • Y • Y NDataPhys.org 2015 [NLC ∗
15] Paper AR • • • • N • N YCHI 2015 [THK ∗
15] Paper • • • • • • Y • N YCHI 2015 [LCN15] Paper • • • • N • N YDataPhys.org 2015 [Bou15] Paper • • • • • • • N • N NCHI 2015 [NB15] Paper • • • • • N • Y YDataPhys.org 2015 [Hei15] Website • • • • N • Y NCHI 2015 [KLA ∗
15] • • • Y • N YDataPhys.org 2015 [MCG ∗
15] Paper • • • • N • N NSIGGRAPH 2015 [ZYZZ15] Paper • • • • • N • Y NCHI 2015 [KLA ∗
15] Paper • • • • Y • N YSIGGRAPH 2015 [DLL ∗
15] Paper • • • • N • N NDataPhys.org 2016 [TL16] Paper • • • • • N • N NSIGGRAPH Art 2016 [Gon16] Paper • • • • Y • Y NSIGGRAPH 2016 [ZLW ∗
16] Paper • • • • N • Y YDataPhys.org 2016 [OPSR16] Short P. • • • • • Y • N NDataPhys.org 2016 [BPAC16] Paper • • • • Y • N YDataPhys.org 2016 [AS16] Website • • • • N • Y NDataPhys.org 2016 [McC16] Website • • • • N • N NDataPhys.org 2016 [KAL ∗
16] Paper • • • • • Y • N YCHI 2016 [HGG ∗
16] Paper • • • • Y • N YCHI 2016 [KGM ∗
16] Paper • • • • N • Y YSIGGRAPH 2016 [MIWI16] Paper Projection • • • Y • N NDataPhys.org 2016 [LGKP ∗
16] Paper • • • • Y • N NDataPhys.org 2016 [iT16] Website • • • • • Y • Y NDataPhys.org 2016 [TACS16] Video • • • • • Y • N NDataPhys.org 2016 [HV16] Short P. • • • • • N • Y NCUMINCAD 2016 [HC17] Short P. • • • • N • N NSIGGRAPH 2016 [SPG ∗
16] Paper • • • • • Y • Y NDataPhys.org 2016 [GD17] Paper • • • • Y • Y YCHI 2016 [SRK ∗
16] Paper • • • • • N • N NTVCG 2016 [SB16] Paper • • • • • • • Y • N NDataPhys.org 2017 [Kan17] Website • • • • N • N NTVCG 2017 [TJW ∗
16] Paper • • • • • Y • N YCHI 2017 [KAP ∗
17] Paper • • • • N • N YCUMINCAD 2017 [PTPM17] Paper • • • • • N • Y NDataPhys.org 2017 [BI18] Paper • • • • Y • N YIEEE CG&A 2017 [DMAS17] Paper • • • • Y • Y NDataPhys.org 2017 [Mei17] Website • • • • • Y • N NEurographics 2018 [BRWM18] Paper • • • • N • Y NCHI 2018 [VKBR ∗
18] Paper • • • • • Y • Y YDataPhys.org 2018 [Kou18] Paper • • • • N • N YDataPhys.org 2018 [ZDX ∗
18] Paper • • • • N • Y YDataPhys.org 2018 [BKW ∗
18] Paper • • • • • • • N • N NEurovis 2018 [ADMAS18] Paper • • • • • Y • Y NDataPhys.org 2018 [Geu18] Paper • • • • • N • Y NCHI 2018 [OTS ∗
18] Paper • • • Y • N NSIGGRAPH Art 2018 [SRP18] Paper • Projection • • • • N • N NCHI 2018 [MTP ∗
18] Paper Projection • • • Y • N YDataPhys.org 2018 [DEBS18] Paper Projection • • • Y • N NSIGGRAPH 2018 [TGZ18] Paper • • • • • N • N NDataPhys.org 2018 [DRC18] Short P. • • • Y • Y NIEEE CG&A 2018 [KJA ∗
18] Paper LED • • • Y • Y YSIGGRAPH Art 2018 [Kat18] Paper • • • • Y • N NCHI 2018 [ZC18] Paper • • • • Y • N NEurographics 2018 [PPW18] Paper • • • • Y • Y YDataPhys.org 2018 [Rod18] Paper • • • • N • N YSIGGRAPH 2018 [LMAH ∗
18] Paper • • • • • Y • N NTVCG 2019 [LGPF ∗
18] Paper • • • • • • • Y • N YDataPhys.org 2019 [ASS ∗
19] Paper • • • • • N • Y YCHI 2019 [NTWVD19] Paper • • • • N • Y NDataPhys.org 2019 [HDA ∗
19] Paper LED • • • • N • Y NEurographics 2019 [TSW ∗
19] Paper • • • • Y • N NACM 2020 [KHM20] Paper • • • • • N • N YCHI 2020 [KHT ∗
20] Paper Projection • • • N • N YDataPhys.org 2020 [MDES21] Paper • • • • Y • Y Y
Table 4: Taxonomy of the works reviewed in this survey (Part 2). © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process
8. Authors’ Short BiographiesHessam Djavaherpour is a Ph.D. Candidate in ComputationalMedia Design (CMD) at the University of Calgary. He is an ar-chitect and a computational designer interested in physical visu-alization of data, data-driven design approaches, algorithmic andparametric design, and digital fabrication. His core Ph.D. publi-cations are mainly focused on pursuing the concept of physical-izing geospatial data at various scales and applications, such asphysicalization of a partial globe [DMAS17], 3D printing land-scapes [ADMAS18], geospatial physyicalization as an urban struc-ture [HDA ∗ Faramarz Samavati is a professor of Computer Science at theUniversity of Calgary. Dr. Samavati’s research interests includeComputer Graphics, Visualization, and Digital Earth. Over the pasteight years, he has received seven best paper awards, Digital Al-berta Award, Great Supervisor Award, University of Calgary PeakAward and Faculty of Science Established Career Scholarship Ex-cellence Award. He has supervised several graduate students withtheses related to physicalization. He has published many papersin this area, including geospatial physicalization [DMAS17, AD-MAS18, MDES21], physicalizing cardiac blood flow data via 3Dprinting [ASS ∗ ∗ Ali Mahdavi-Amiri is currently a University Research Associatein the Department of Computing Science at Simon Fraser Univer-sity. His primary research interest is in visual computing with focuson geometry processing, computational fabrication, and machinelearning. His work on computational fabrication covers a varietyof research problems and fabrication techniques such as machin-ing optimization (3-axis CNC) [MAYZ ∗ ∗ ∗ Fatemeh Yazdanbakhsh is a Ph.D. student at the University ofCalgary working on the application of physicalization for produc-ing medical prototypes. Her research focus is on making physi-cal models as a replacement for cadaveric bones used for temporalbone surgery rehearsal and teaching purposes. She explores differ-ent materials to find the best match for reproducing tactile sense forhard and soft tissue. Her research also includes finding a method tofabricate complicated structures in different colours and materialsusing off-the-shelf 3D printers.
Samuel Huron is an associate professor in Design of Informa-tion Technologies inside the Social and Economical Science De-partment of Telecom Paris School at the Institut Polytechnique deParis, and part of the CNRS Institut Interdisciplinaire of innova-tion. He is deeply interested in how the construction process ofphysical representation of data impact the cognitive process of theauthors. He has worked extensively on the topic of constructingdata physicalisation, as a design paradigm [HCT ∗ ∗ ∗ ∗
17] and also ina multitude of non-academic setting.
Richard Levy has recently retired from the University of Calgarywhere he was a Professor of Planning and Urban Design at TheUniversity of Calgary for 26 years. Dr. Levy also served as theCo-Director of the Computational Media Program (CMD) and isan Adjunct Professor in the Department of Computer Science andThe Department of Archaeology. Dr. Levy has conducted researchprojects with faculty from Archaeology, Computer Science, Geo-matics Engineering, Kinesiology and Psychology. Dr. Levy speaksat international and national conferences in the fields of archaeol-ogy, education, serious games, urban planning, and virtual reality.
Yvonne Jansen is a tenured research scientist with the FrenchNational Center for Scientific Research (CNRS) and a memberof the Institute for Intelligent Systems and Robotics at SorbonneUniversité. She has published extensively on the topic of dataphysicalization including a handbook chapter [DJVM21], a reviewand research agenda [JDI ∗ ∗ Lora Oehlberg is an Associate Professor of Computer Scienceat the University of Calgary. Her background is in design theoryin methodology, which she applies to the design of technologiesto support creativity and collaboration. She has published severalpapers at the intersection of physical authoring tools and data vi-sualization, most notably using physical graphical template toolsfor [WOSC19], considering alternative fabrication media for dataphysicalization [WWOC19] and proposing a physicalization au-thoring tool with Yvonne Jansen [SSJ ∗ © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process References [ADMAS18] A
LLAHVERDI
K., D
JAVAHERPOUR
H., M
AHDAVI -A MIRI
A., S
AMAVATI
F.: Landscaper: A modeling system for 3D printing scalemodels of landscapes.
Computer Graphics Forum 37 , 3 (2018), 439–451.1, 7, 11, 12, 14, 17, 18, 20, 21[AFS05] A
ITCHESON
R., F
RIEDMAN
J., S
EEBOHM
T.: 3-axis cncmilling in architectural design.
International Journal of ArchitecturalComputing 3 , 2 (2005), 161–180. 1, 13[AS16] A
LLEN
B., S
MITH
S.: Motus forma: People’s motions in a sharedspace, 2016. Last Accessed Jan 2021. URL: http://dataphys.org/list/motus-forma-peoples-motions-in-a-shared-space/. 5, 7, 8, 9, 14, 18, 20[ASS ∗
19] A NG K. D., S
AMAVATI
F. F., S
ABOKROHIYEH
S., G
ARCIA
J., E
LBAZ
M. S.: Physicalizing cardiac blood flow data via 3D printing.
Computers & Graphics 85 (2019), 42–54. 3, 4, 5, 6, 12, 17, 20, 21[Ass21] A
SSOCIATES
UTODESK
UTODESK
WEIDA
C.: Robot arranges 8,000 nails into a data sculpture ofthe wind, 2013. Last Accessed Jan 2021. URL: http://dataphys.org/list/robot-arranges-8000-nails-into-a-data-sculpture-of-the-wind/. 7, 10,16, 17, 19[Bad13] B
ADGER
E.: Population density emerging from walls,2013. Last Accessed Jan 2021. URL: http://dataphys.org/list/population-density-emerging-from-walls/. 7, 19[Bar11] B
ARRASS
S.: Phsyical sonification dataforms. In
Proceedingsof the 17th International Conference on Auditory Display (ICAD2011) (2011), International Community for Auditory Display. 6, 9, 17, 19[Bar12] B
ARRASS
S.: Digital fabrication of acoustic sonifications.
Jour-nal of the Audio Engineering Society 60 , 9 (2012), 709–715. 6, 9, 12,19[Bar14] B
ARRASS
S.: Acoustic sonification of blood pressure in the formof a singing bowl. In
Conference on Sonification of Health and Environ-mental Data (SoniHED) (2014). 6, 12, 19[BFR17] B
ERMANO
A. H., F
UNKHOUSER
T., R
USINKIEWICZ
S.: Stateof the art in methods and representations for fabrication-aware design.
Computer Graphics Forum 36 , 2 (2017), 509–535. 2[BH12] B
ROWN
C., H
URST
A.: Viztouch: automatically generated tac-tile visualizations of coordinate spaces. In
Proceedings of the Sixth Inter-national Conference on Tangible, Embedded and Embodied Interaction (2012), Association for Computing Machinery (ACM), pp. 131–138. 6,12, 19[BHR ∗
17] B
ÜCKING
T. M., H
ILL
E. R., R
OBERTSON
J. L., M
ANEAS
E., P
LUMB
A. A., N
IKITICHEV
D. I.: From medical imaging data to3D printed anatomical models.
PloS one 12 , 5 (2017), e0178540. 1[BI18] B
OEM
A., I
WATA
H.: “it’s like holding a human heart”: the designof vital+ morph, a shape-changing interface for remote monitoring.
AI& SOCIETY 33 , 4 (2018), 599–619. 20[BKW ∗
18] B
ADER
C., K
OLB
D., W
EAVER
J. C., S
HARMA
S., H
OSNY
A., C
OSTA
J., O
XMAN
N.: Making data matter: Voxel printing for thedigital fabrication of data across scales and domains.
Science advances4 , 5 (2018), eaas8652. 4, 5, 6, 13, 17, 20[BL12] B
ARRY
D., L
ASKOWITZ
A.: Saturation. In
ACM SIGGRAPH2012 Art Gallery (New York, NY, USA, 2012), Association for Comput-ing Machinery (ACM), p. 374–375. 8, 17, 19[BM17] B
ROWN
N., M
UELLER
C.: Designing with data: moving beyondthe design space catalog. In “Disciplines and Disruption" Proceedings of the 37th Annual Conference of the Association for Computer AidedDesign in Architecture (ACADIA) (2017), CUMINCAD, pp. 154–163. 4[Bou15] B
OURKE
P.: Novel physical representations for the visualisa-tion of science data and mathematics. In
International Conference onComputer Games, Multimedia \ &Allied Technology (CGAT). Proceed-ings (2015), p. 13. 6, 20[Bow11] B OWEN
D.: David bowen’s sea wave replicators,2011. Last Accessed Jan 2021. URL: http://dataphys.org/list/tele-present-water-and-underwater-sea-wave-replication/. 19[BPAC16] B
OTROS
F., P
ERIN
C., A
SENIERO
B. A., C
ARPENDALE
S.:Go and grow: Mapping personal data to a living plant. In
Proceedingsof the International Working Conference on Advanced Visual Interfaces (2016), Association for Computing Machinery (ACM), pp. 112–119. 7,20[BRWM18] B
IRSAK
M., R
IST
F., W
ONKA
P., M
USIALSKI
P.: Stringart: towards computational fabrication of string images.
ComputerGraphics Forum 37 , 2 (2018), 263–274. 8, 16, 20[BSJ98] B
AILEY
M. J., S
CHULTEN
K., J
OHNSON
J. E.: The use ofsolid physical models for the study of macromolecular assembly.
CurrentOpinion in Structural Biology 8 , 2 (1998), 202–208. 3, 12, 19[CB17] C
ARROLL
F. A., B
LAUCH
D. N.: 3D printing of molecularmodels with calculated geometries and p orbital isosurfaces.
Journalof Chemical Education 94 , 7 (2017), 886–891. 1[CO14] C
HEN
X., O
LIVEIRA
P. G. C.: x.pose: a wearable dynamicdata sculpture, 2014. Last Accessed Jan 2021. URL: http://xc-xd.com/x-pose. 5, 8, 19[DEBS18] D
ADKHAHFARD
S., E
TEMAD
K., B
ROSZ
J., S
AMAVATI
F.:Area preserving dynamic geospatial visualization on physical globe. In
VISIGRAPP (3: IVAPP) (2018), pp. 309–318. 7, 12, 14, 17, 18, 20[DeM11] D E M ARCO
P.: Paper models of 3D plots, 2011. Last AccessedJan 2021. URL: http://dataphys.org/list/paper-models-of-3d-plots/. 6,19[DJVM21] D
RAGICEVIC
P., J
ANSEN
Y., V
ANDE M OERE
A.: Data Phys-icalization. In
Springer Handbook of Human Computer Interaction ,Springer Reference. Springer, 2021. 21[DLL ∗
15] D
UMAS
J., L U A., L
EFEBVRE
S., W U J., D
ICK
C.: By-example synthesis of structurally sound patterns.
ACM Transactions onGraphics (TOG) 34 , 4 (2015), 1–12. 8, 20[DMAS17] D
JAVAHERPOUR
H., M
AHDAVI -A MIRI
A., S
AMAVATI
F. F.: Physical visualization of geospatial datasets.
IEEE computergraphics and applications 37 , 3 (2017), 61–69. 1, 4, 7, 10, 14, 17, 18,20, 21[DRC18] D
ANIEL
M., R
IVIÈRE
G., C
OUTURE
N.: Designing an ex-pandable illuminated ring to build an actuated ring chart. In
Proceed-ings of the Twelfth International Conference on Tangible, Embedded,and Embodied Interaction (2018), Association for Computing Machin-ery (ACM), pp. 140–147. 15, 16, 20[Dun12] D
UNN
N.:
Digital fabrication in architecture . Laurence King,London, 2012. 9, 10, 11, 13[Dwy04] D
WYER
T.:
Two-and-a-half-dimensional Visualisation of Rela-tional Networks . PhD thesis, University of Sydney, 2004. 3[Dwy05] D
WYER
T.: Time-evolving scatterplot, 2005. Last AccessedJan 2021. URL: http://dataphys.org/list/time-evolving-scatterplot/. 13,19[EA17] E
VERITT
A., A
LEXANDER
J.: Polysurface: a design approachfor rapid prototyping of shape-changing displays using semi-solid sur-faces. In
Proceedings of the 2017 Conference on Designing Inter-active Systems (2017), Association for Computing Machinery (ACM),pp. 1283–1294. 13[Epl12] E
PLER
M.: Grand old party: Political satire, 2012.Last Accessed Jan 2021. URL: http://dataphys.org/list/grand-old-party-political-satire/. 8, 10, 19 © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process [FF14] F
ENS
P., F
UNK
M.: Personal health data: visualization modal-ities and their perceived values. In (2014), VáclavSkala-UNION Agency, pp. 339–344. 4, 5, 6, 13, 20[FLO ∗
13] F
OLLMER
S., L
EITHINGER
D., O
LWAL
A., H
OGGE
A., I
SHII
H.: inform: dynamic physical affordances and constraints through shapeand object actuation. In
Proceedings of the 26th Annual ACM Symposiumon User Interface Software and Technology (2013), vol. 13, Associationfor Computing Machinery, pp. 417–426. 4, 15, 16[FM12] F
ERRARA
C., M C T AGGERT
J.: Pulse: Tangible line graph,2012. Last Accessed Jan 2021. URL: http://dataphys.org/list/pulse-tangible-line-graph/. 6, 19[Fou21] F
OUNDATION RY B., R
EAS
C.: A short introduction to the processing softwareand projects from the community., 2020. Last Accessed Dec 2020. URL:https://processing.org/overview/. 10, 11[FWF ∗
14] F
ETTERMAN
M. R., W
EBER
Z. J., F
REKING
R. A., V
OLPE
A., S
COTT
D.: Luminocity: a 3D printed, illuminated city generatedfrom ladar data. In (2014), IEEE, pp. 1–4. 1, 18[GD17] G
OURLET
P., D
ASSÉ
T.: Cairn: A tangible apparatus for situ-ated data collection, visualization and analysis. In
Proceedings of the2017 Conference on Designing Interactive Systems (2017), Associationfor Computing Machinery (ACM), pp. 247–258. 8, 9, 13, 14, 18, 20[Geu18] G
EURTS
K.: Imagining thought in digital space: 3D printedthoughts. In
Proceedings of the 5th Biennial Transdisciplinary Imag-ing Conference (2018), Biennial Transdisciplinary Imaging Conference,pp. 81–99. 5, 12, 20[GG15] G
EURTS
K., G
UGLIELMETTI
M.: Imagining thought in digitalspace. In
Procedding of the 21th International Symposium on ElectronicArt ISEA2015 (2015), ISEA: International Symposium on Electronic Art,pp. 1–4. 5[GHHS14] G
EIGER
C., H
OGEN
M., H
ORNIG
J., S
CHAAR
M.: Daten-reise: digital bits made tangible. In
Proceedings of the 8th InternationalConference on Tangible, Embedded and Embodied Interaction (2014),Association for Computing Machinery (ACM), pp. 283–284. 8, 19[GHK12] G
OODS
D., H
AFERMAAS
N., K
OBLIN
A.: ecloud & airfield:Ambient airport visualizations, 2012. Last Accessed Jan 2021. URL:http://dataphys.org/list/ecloud-airfield-ambient-airport-visualizations/.8, 19[GI18] G
RUNDHÖFER
A., I
WAI
D.: Recent advances in projection map-ping algorithms, hardware and applications.
Computer Graphics Forum37 , 2 (2018), 653–675. 14[Gon16] G
ONI
K.: Deletion process_only you can see my history: in-vestigating digital privacy, digital oblivion, and control on personal datathrough an interactive art installation. In
ACM SIGGRAPH 2016 ArtGallery . Association for Computing Machinery (ACM), 2016, pp. 324–333. 8, 11, 20[GR17] G
ONZALEZ R OJAS
P.: Space and motion: Data-driven model of4d pedestrian behavior. In “Disciplines and Disruption" Proceedingsof the 37th Annual Conference of the Association for Computer AidedDesign in Architecture (ACADIA) (2017), CUMINCAD, pp. 266–273. 4[GSSO05] G
ILLET
A., S
ANNER
M., S
TOFFLER
D., O
LSON
A.: Tangi-ble interfaces for structural molecular biology.
Structure 13 , 3 (2005),483–491. 3, 4, 5, 12, 15, 17, 19[GWW ∗
04] G
ILLET
A., W
EGHORST
S., W
INN
W., S
TOFFLER
D.,S
ANNER
M., G
OODSELL
D., O
LSON
A.: Computer-linked autofabri-cated 3D models for teaching structural biology. In
ACM SIGGRAPH2004 Sketches . Association for Computing Machinery (ACM), 2004,p. 28. 5, 15, 19[GYS ∗
12] G
WILT
I., Y
OXALL
A., S
ANO
K.,
ET AL .: Enhancing theunderstanding of statistical data through the creation of physical objects. In
DS 73-1 Proceedings of the 2nd International Conference on DesignCreativity Volume 1 (2012), pp. 117–126. 4, 8, 12, 17, 19[Gü11] G
ÜNGÖR
C.: Blip: A year of travel, 2011. Last Accessed Jan2021. URL: http://dataphys.org/list/blip-a-year-of-travel/. 8, 9, 13, 19[HAA ∗
20] H
OSSEINI
S. V., A
LIM
U. R., A
MIRI
A. M., O
EHLBERG
L., T
ARON
J. M.: Portal: Design and fabrication of incidence-drivenscreens. In
The Proceedings of the SMI 2020 Fabrication & SculptingEvent (FASE) (2020), HYPERSEEING, p. 31. 8, 10, 21[HAD ∗
18] H
ADEED
K., A
CAR
P., D
ULAC
Y., C
UTTONE
F., A LA - COQUE
X., K
ARSENTY
C.: Cardiac 3D printing for better understandingof congenital heart disease.
Archives of cardiovascular diseases 111 , 1(2018), 1. 1[HC17] H
UANG
A., C
HANEY
A.: Data moiré: Optical patterns as data-driven design narratives. In “Disciplines and Disruption" Proceedingsof the 37th Annual Conference of the Association for Computer AidedDesign in Architecture (ACADIA) (2017), ACADIA, pp. 162–167. 6, 10,13, 14, 17, 20[HCT ∗
14] H
URON
S., C
ARPENDALE
S., T
HUDT
A., T
ANG
A.,M
AUERER
M.: Constructive visualization. In
Proceedings of the 2014conference on Designing interactive systems (2014), Association forComputing Machinery (ACM), pp. 433–442. 18, 21[HDA ∗
19] H
OSSEINI
S. V., D
JAVAHERPOUR
H., A
LIM
U. R., T
ARON
J. M., S
AMAVATI
F. F.: Data-spatialized pavilion: Introducing a data-driven design method based on principles of catoptric anamorphosis. In
International Society of the Arts, Mathematics, and Architecture, Sum-mer 2019, The Proceedings of the SMI 2019 Fabrication and SculptungEvent (2019), pp. 39–51. 5, 7, 10, 12, 17, 20, 21[Hei15] H
EINICKER
P.: Data necklace of good night sms,2015. Last Accessed Jan 2021. URL: http://dataphys.org/list/data-necklace-of-good-night-sms/. 8, 12, 20[Hem13] H
EMMENT
D.: Emoto-visualising the online response to lon-don 2012. In
ISEA International; Australian Network for Art & Technol-ogy; University of Sydney (2013), ISEA International; Australian Net-work for Art & Technology; University of Sydney. 4, 6, 14, 17, 19[HGG ∗
16] H
OUBEN
S., G
OLSTEIJN
C., G
ALLACHER
S., J
OHNSON
R.,B
AKKER
S., M
ARQUARDT
N., C
APRA
L., R
OGERS
Y.: Physikit: Dataengagement through physical ambient visualizations in the home. In
Proceedings of the 2016 CHI Conference on Human Factors in Com-puting Systems (2016), Association for Computing Machinery (ACM),pp. 1608–1619. 4, 8, 13, 20[HGH ∗
17] H
URON
S., G
OURLET
P., H
INRICHS
U., H
OGAN
T.,J
ANSEN
Y.: Let’s get physical: Promoting data physicalization in work-shop formats. In
Proceedings of the 2017 Conference on Designing In-teractive Systems (2017), Association for Computing Machinery (ACM),pp. 1409–1422. 21[HHA ∗
18] H
OGAN
T., H
INRICHS
U., A
LEXANDER
J., H
URON
S.,C
ARPENDALE
S., H
ONECKER
E.: Toward a design language for dataphysicalization. In
IEEE VIS, Oct 2018 (2018). 21[HHJ ∗
17] H
OGAN
T., H
INRICHS
U., J
ANSEN
Y., H
URON
S., G
OURLET
P., H
ORNECKER
E., N
ISSEN
B.: Pedagogy & physicalization: Design-ing learning activities around physical data representations. In
Proceed-ings of the 2017 ACM Conference Companion Publication on Design-ing Interactive Systems (2017), Association for Computing Machinery(ACM), pp. 345–347. 21[HIH ∗
13] H
ULLIN
M. B., I
HRKE
I., H
EIDRICH
W., W
EYRICH
T.,D
AMBERG
G., F
UCHS
M.: Computational Fabrication and Display ofMaterial Appearance. In
Eurographics 2013 - State of the Art Reports (2013), Sbert M., Szirmay-Kalos L., (Eds.), The Eurographics Associa-tion (ACM). 2, 11[HJC14] H
URON
S., J
ANSEN
Y., C
ARPENDALE
S.: Constructing visualrepresentations: Investigating the use of tangible tokens.
IEEE transac-tions on visualization and computer graphics 20 , 12 (2014), 2102–2111.1, 21 © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process [HKH ∗
04] H
OLSTIUS
D., K
EMBEL
J., H
URST
A., W AN P.-H., F OR - LIZZI
J.: Infotropism: living and robotic plants as interactive displays.In
Proceedings of the 5th conference on Designing interactive systems:processes, practices, methods, and techniques (2004), Association forComputing Machinery (ACM), pp. 215–221. 8, 19[HSBH17] H
IGMAN
C. S., S
ITU
H., B
LACKLIN
P., H
EIN
J. E.: Hands-on data analysis: Using 3D printing to visualize reaction progress sur-faces.
Journal of Chemical Education 94 , 9 (2017), 1367–1371. 1[Hus14] H
USH : Made by numbers, 2014. Last Accessed Jan 2021. URL:http://dataphys.org/list/made-by-numbers/. 13, 17, 19[HV16] H
ÄKKILÄ
J., V
IRTANEN
L.: Aesthetic physical items for vi-sualizing personal sleep data. In
Proceedings of the 18th InternationalConference on Human-Computer Interaction with Mobile Devices andServices Adjunct (2016), Association for Computing Machinery (ACM),pp. 1173–1177. 7, 9, 10, 20[HW17] H
ULL
C., W
ILLETT
W.: Building with data: Architectural mod-els as inspiration for data physicalization. In
Proceedings of the 2017CHI Conference on Human Factors in Computing Systems (2017), As-sociation for Computing Machinery (ACM), pp. 1217–1264. 9[HW18] H
ULL
C. H., W
ILLETT
W.: Data tectonics: A framework forbuilding physical and immersive data representations.
IEEE ComputerGraphics and Applications 38 , 5 (2018), 11–17. 9[Inc03] I NC . S. T. M.: Solid terrain model with airplane trajec-tory, 2003. Last Accessed Jan 2021. URL: http://dataphys.org/list/solid-terrain-model-with-airplane-trajectory/. 4, 7, 14, 19[Ins16] I NSTITUTION
S.: Blombos ocher plaque, 2016. Last Ac-cessed Dec 2020. URL: http://humanorigins.si.edu/evidence/behavior/recording-information/blombos-ocherplaque. 1[iT16] I S KETCH L AB , T RACK M AVEN : Podium: Physical competitionmonitor, 2016. Last Accessed Jan 2021. URL: http://dataphys.org/list/podium-physical-competition-monitor/. 8, 20[Jab13] J
ABI
W.:
Parametric design for architecture . Laurence King,London, 2013. 10[Jan14] J
ANSEN
Y.:
Physical and tangible information visualization .PhD thesis, Université Paris Sud-Paris XI, 2014. 4[JD13] J
ANSEN
Y., D
RAGICEVIC
P.: An interaction model for visual-izations beyond the desktop.
IEEE Transactions on Visualization andComputer Graphics (TVCG) 19 , 12 (2013), 2396–2405. 1, 4, 9, 11, 21[JDF13] J
ANSEN
Y., D
RAGICEVIC
P., F
EKETE
J.-D.: Evaluating the ef-ficiency of physical visualizations. In
Proceedings of the SIGCHI Con-ference on Human Factors in Computing Systems (2013), Association forComputing Machinery (ACM), pp. 2593–2602. 1, 4, 6, 13, 18, 19, 21[JDI ∗
15] J
ANSEN
Y., D
RAGICEVIC
P., I
SENBERG
P., A
LEXANDER
J.,K
ARNIK
A., K
ILDAL
J., S
UBRAMANIAN
S., H
ORNBÆK
K.: Oppor-tunities and challenges for data physicalization. In
Proceedings of the33rd Annual ACM Conference on Human Factors in Computing Systems (2015), Association for Computing Machinery (ACM), pp. 3227–3236.1, 21[JG14] J
ARMAN
R., G
ERHARDT
J.: Cosmos: Carbon exchange capturedin a wooden ball, 2014. Last Accessed Jan 2021. URL: http://dataphys.org/list/cosmos-carbon-emission-captured-in-a-wooden-ball/. 6, 13, 20[JS11] J
ORGE
J., S
AMAVATI
F.:
Sketch-based Interfaces and Modeling .Springer London, 2011. 10[KAL ∗
16] K
HOT
R. A., A
NDRES
J., L AI J., V ON K AENEL
J.,M
UELLER
F.: Fantibles: Capturing cricket fan’s story in 3D. In
Pro-ceedings of the 2016 ACM Conference on Designing Interactive Systems (2016), Association for Computing Machinery (ACM), pp. 883–894. 6,8, 12, 18, 20[Kan17] K
ANG
J.: Wearable self, 2017. Last Accessed Jan 2021. URL:http://dataphys.org/list/wearable-self/. 5, 7, 14, 20[KAP ∗
17] K
HOT
R. A., A
GGARWAL
D., P
ENNINGS
R., H
JORTH
L.,M
UELLER
F.: Edipulse: investigating a playful approach to self-monitoring through 3D printed chocolate treats. In
Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (2017),Association for Computing Machinery (ACM), pp. 6593–6607. 5, 7, 20[Kat18] K
ATSUMOTO
Y.: Robotype: studies of kinetic typography byrobot display for expressing letters, time and movement.
Leonardo 51 , 4(2018), 386–393. 8, 9, 10, 14, 20[KB13] K
AUFFMAN
B., B
RENNER
S.: NYC high school dropouts,2013. Last Accessed Dec 2020. URL: http://dataphys.org/list/nyc-high-school-dropouts/. 4, 6, 12, 19[KB14] K
ANE
S. K., B
IGHAM
J. P.: Tracking@ stemxcomet: teachingprogramming to blind students via 3D printing, crisis management, andtwitter. In
Proceedings of the 45th ACM technical symposium on Com-puter science education (2014), Association for Computing Machinery(ACM), pp. 247–252. 4, 7, 12, 17, 19[Kel09] K
ELLER
C.: Datamorphose: Animated sails, 2009.Last Accessed Jan 2021. URL: http://dataphys.org/list/displaying-data-through-animated-sails/. 8, 16, 19[KGM ∗
16] K
AZI
R. H., G
ROSSMAN
T., M
OGK
C., S
CHMIDT
R., F
ITZ - MAURICE
G.: Chronofab: Fabricating motion. In
Proceedings of the2016 CHI Conference on Human Factors in Computing Systems (2016),Association for Computing Machinery (ACM), pp. 908–918. 8, 13, 20[Kha10] K
HABAZI
Z.:
Generative algorithms using grasshopper ,vol. 3. Digitally published, 2010. Last Accessed Jan 2021.URL: http://download.mcneel.com/s3/mcneel/grasshopper/1.0/docs/en/Generative%20Algorithms.pdf. 9[KHM14] K
HOT
R. A., H
JORTH
L., M
UELLER
F.: Understanding phys-ical activity through 3D printed material artifacts. In
Proceedings of theSIGCHI Conference on Human Factors in Computing Systems (2014),Association for Computing Machinery (ACM), pp. 3835–3844. 5, 6, 12,17, 19[KHM20] K
HOT
R. A., H
JORTH
L., M
UELLER
F.: Shelfie: A frameworkfor designing material representations of physical activity data.
ACMTransactions on Computer-Human Interaction (TOCHI) 27 , 3 (2020),1–52. 7, 12, 20[KHT ∗
20] K
IRSHENBAUM
N., H
UTCHISON
J., T
HERIOT
R.,K
OBAYASHI
D., L
EIGH
J.: Data in context: Engaging audienceswith 3D physical geo-visualization. In
Extended Abstracts of the 2020CHI Conference on Human Factors in Computing Systems (2020),Association for Computing Machinery (ACM), pp. 1–9. 3, 4, 12, 17, 20[Kil14] K
ILDALL
S.: Water works: Maps of the san francisco water in-frastructure, 2014. Last Accessed Jan 2021. URL: http://dataphys.org/list/water-works/. 7, 19[Kis09] K
ISON
M.: Pulse: Animated heart shows sentiments,2009. Last Accessed Jan 2021. URL: http://dataphys.org/list/pulse-showing-emotional-responses-on-the-web/. 8, 19[KJA ∗
18] K
EEFE
D. F., J
OHNSON
S., A
LTHEIMER
R., H
ONG
D.-G.,H
UNTER
R., J
OHNSON
A. J., R
OCKCASTLE
M., S
WACKHAMER
M.,W
ITTKAMPER
A.: Weather report: A site-specific artwork interweavinghuman experiences and scientific data physicalization.
IEEE ComputerGraphics and Applications 38 , 4 (2018), 10–16. 5, 8, 17, 20[KLA ∗
15] K
HOT
R. A., L EE J., A
GGARWAL
D., H
JORTH
L.,M
UELLER
F.: Tastybeats: Designing palatable representations of phys-ical activity. In
Proceedings of the 33rd Annual ACM Conference onHuman Factors in Computing Systems (2015), Association for Comput-ing Machinery (ACM), pp. 2933–2942. 7, 20[Kna12] K
NAPEK
M.: Windcuts: Wind travels captured on wood,2012. Last Accessed Jan 2021. URL: http://dataphys.org/list/windcuts-wind-travels-captured-on-wood/. 8, 13, 14, 19[Kol04] K
OLAREVIC
B.:
Architecture in the digital age: design and man-ufacturing . taylor & Francis, 2004. 11, 13[Kos07] K
OSARA
R.: Visualization criticism-the missing link betweeninformation visualization and art. In (2007), IEEE, pp. 631–636. 3 © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process [Kou18] K
OUTSOMICHALIS
M.: Objektivisering: Text physicalizationand self-introspective post-digital objecthood. In
Proceedings of theTwelfth International Conference on Tangible, Embedded, and Embod-ied Interaction (2018), Association for Computing Machinery (ACM),pp. 521–528. 8, 12, 20[KS12] K. S
CHNEIDER
D.: Thesis lego board, 2012. Last Accessed Jan2021. URL: http://dataphys.org/list/thesis-lego-board/. 4, 8, 9, 12, 18,19[KS13] K
NILL
O., S
LAVKOVSKY
E.: Illustrating mathematics using 3Dprinters. arXiv:1306.5599 . 1[LBRM12] L UO L., B
ARAN
I., R
USINKIEWICZ
S., M
ATUSIK
W.:Chopper: Partitioning models into 3D-printable parts.
ACM Transac-tions on Graphics (TOG) 31 , 6 (November 2012). 12, 18[LCN15] L EE M.-H., C HA S., N AM T.-J.: Patina engraver: Visualizingactivity logs as patina in fashionable trackers. In
Proceedings of the33rd Annual ACM Conference on Human Factors in Computing Systems (2015), Association for Computing Machinery (ACM), pp. 1173–1182.5, 7, 17, 20[LEM ∗
17] L
IVESU
M., E
LLERO
S., M
ARTÍNEZ
J., L
EFEBVRE
S., A T - TENE
M.: From 3D models to 3D prints: an overview of the processingpipeline.
Computer Graphics Forum 36 , 2 (2017), 537–564. 2, 12, 13[Len12] L
ENG ∗
16] L E G OC M., K IM L. H., P
ARSAEI
A., F
EKETE
J.-D.,D
RAGICEVIC
P., F
OLLMER
S.: Zooids: Building blocks for swarm userinterfaces. In
Proceedings of the 29th Annual Symposium on User Inter-face Software and Technology (2016), Association for Computing Ma-chinery (ACM), pp. 97–109. 8, 16, 20[LGPF ∗
18] L E G OC M., P
ERIN
C., F
OLLMER
S., F
EKETE
J.-D.,D
RAGICEVIC
P.: Dynamic composite data physicalization usingwheeled micro-robots.
IEEE transactions on visualization and computergraphics 25 , 1 (2018), 737–747. 4, 6, 8, 16, 20[LH04a] L
OZANO -H EMMER
R.: 2004 – synaptic caguamas: Visualizecellular automata, 2004. Last Accessed Jan 2021. URL: http://dataphys.org/list/synaptic-caguamas-visualize-cellular-automata/. 5, 19[LH04b] L
OZANO -H EMMER
R.: Array of belts visualize public pres-ence, 2004. Last Accessed Jan 2021. URL: http://dataphys.org/list/array-of-belts-visualize-public-presence/. 8, 15, 17, 19[LH06] L
OZANO -H EMMER
R.: Pulse room: Light bulbs show heartbeats, 2006. Last Accessed Jan 2021. URL: http://dataphys.org/list/pulse-room-visualize-detected-heart-beats/. 5, 19[LH07] L
OZANO -H EMMER
R.: Wavefunction: Visualize public presencein waves, 2007. Last Accessed Jan 2021. URL: http://dataphys.org/list/wavefunction-visualize-public-presence-in-waves/. 6, 19[LH11] L
OZANO -H EMMER
R.: Tape recorders: Time-measuring tapes,2011. Last Accessed Jan 2021. URL: http://dataphys.org/list/tape-recorders-measure-time-instead-of-distance/. 6, 7, 15, 17, 19[LH12] L
OZANO -H EMMER
R.: Pulse drip: Visualize heartbeats with awater hose, 2012. Last Accessed Jan 2021. URL: http://dataphys.org/list/pulse-drip-visualize-heartbeats-with-a-water-hose/. 19[LH14] L
OZANO -H EMMER
R.: Sphere packing: Visu-alize composers’ total musical production, 2014. LastAccessed Jan 2021. URL: http://dataphys.org/list/sphere-packing-visualize-composers-total-musical-production/. 19[LJL12] L
UKASSEN
T., J
ENSEN
H. H., L
INDEGAARD
J. B.:Chaotic flow: Abstract flow visualization of copenhagen bike traf-fic, 2012. Last Accessed Jan 2021. URL: http://dataphys.org/list/chaotic-flow-abstract-flow-visualization-of-copenhagen-bike-traffic/.6, 19[LLD ∗
11] L
EITHINGER
D., L
AKATOS
D., D E V INCENZI
A., B
LACK - SHAW
M., I
SHII
H.: Direct and gestural interaction with relief: a 2.5d shape display. In
Proceedings of the 24th annual ACM symposium onUser interface software and technology (2011), Association for Comput-ing Machinery (ACM), pp. 541–548. 15, 19 [LMAH ∗
18] L I S., M
AHDAVI -A MIRI
A., H U R., L IU H., Z OU C.,V AN K AICK
O., L IU X., H
UANG
H., Z
HANG
H.: Construction andfabrication of reversible shape transforms.
ACM Transactions on Graph-ics (TOG) 37 , 6 (2018), 1–14. 8, 10, 12, 20, 21[LPI11] L EE J., P
OST
R., I
SHII
H.: Zeron: Mid-air tangible interactionenabled by computer controlled magnetic levitation. In
Proceedings ofthe 24th Annual ACM Symposium on User Interface Software and Tech-nology (New York, NY, USA, 2011), UIST ’11, Association for Com-puting Machinery, p. 327–336. URL: https://doi.org/10.1145/2047196.2047239, doi:10.1145/2047196.2047239 . 16[MAAS15] M
AHDAVI -A MIRI
A., A
LDERSON
T., S
AMAVATI
F.: A sur-vey of digital earth.
Computers & Graphics 53 (2015), 95–117. 7[Mad95] M
ADSEN
L.: Loren madsen’s early data sculptures, 1995. LastAccessed Jan 2021. URL: http://dataphys.org/list/cpi-cost-of-living/. 5,6, 13, 19[Mar14] M
ARCUS
A.: Centennial chromagraph: Data spatialization andcomputational craft. In
Proceedings of the 34th Annual Conference ofthe Association for Computer Aided Design in Architecture (ACADIA) (2014), CUMINCAD. 5, 6, 8, 10, 13, 14, 17, 19[MAWS15] M
AHDAVI -A MIRI
A., W
HITTINGHAM
P., S
AMAVATI
F.:Cover-it: an interactive system for covering 3D prints. In
Proceedingsof the 41st Graphics Interface Conference (2015), Canadian InformationProcessing Society, pp. 73–80. 11, 21[MAYZ ∗
20] M
AHDAVI -A MIRI
A., Y U F., Z
HAO
H., S
CHULZ
A.,Z
HANG
H.: Vdac: Volume decompose-and-carve for subtractive manu-facturing.
ACM Transactions on Graphics (TOG) 39 , 6 (2020), 203:1–203:15. 13, 21[McC13] M C C UNE
D.: Doug mccune’s physical maps,2013. Last Accessed Jan 2021. URL: http://dataphys.org/list/doug-mccunes-physical-maps/. 5, 6, 12, 19[McC16] M C C UNE
D.: Sculpture of housing prices rip-ping san francisco apart, 2016. Last Accessed Dec2020. URL: http://dougmccune.com/blog/2016/07/21/sculpture-of-housing-prices-ripping-san-francisco-apart/. 5, 12,20[MCG ∗
15] M
ADURA
T. I., C
LEMENTEL
N., G
ULL
T. R., K
RUIP
C. J.,P
AARDEKOOPER
J.-P.: 3D printing meets computational astrophysics:deciphering the structure of η carinae’s inner colliding winds. MonthlyNotices of the Royal Astronomical Society 449 , 4 (2015), 3780–3794. 8,12, 20[MDES21] M
OORMAN
L., D
JAVAHERPOUR
H., E
TEMAD
K., S
AMA - VATI
F. F.: Geospatial physicalization in geography education.
Journalof Geography 120 , 1 (2021), 23–35. 7, 10, 12, 14, 18, 20, 21[Mei17] M
EIER
S.: Green berlin, 2017. Last Accessed Jan 2021. URL:http://dataphys.org/list/green-berlin/. 7, 13, 14, 20[MIWI16] M
IYASHITA
L., I
SHIHARA
K., W
ATANABE
Y., I
SHIKAWA
M.: Zoematrope: A system for physical material design. In
ACM SIG-GRAPH 2016 Emerging Technologies . Association for Computing Ma-chinery (ACM), 2016, pp. 1–1. 4, 8, 20[MP09] M
OERE
A. V., P
ATEL
S.: The physical visualization of infor-mation: designing data sculptures in an educational context. In
Visualinformation communication . Springer, 2009, pp. 1–23. 8, 19[MTP ∗
18] M
ILLAR
G. C., T
ABRIZIAN
P., P
ETRASOVA
A., P
ETRAS
V.,H
ARMON
B., M
ITASOVA
H., M
EETENMEYER
R. K.: Tangible land-scape: A hands-on method for teaching terrain analysis. In
Proceedingsof the 2018 CHI Conference on Human Factors in Computing Systems (2018), Association for Computing Machinery (ACM), pp. 1–12. 7, 8,14, 17, 20[Mun08] M
UNZNER
T.: Process and pitfalls in writing information visu-alization research papers. In
Information visualization . Springer, 2008,pp. 134–153. 3[Mun14] M
UNZNER
T.:
Visualization analysis and design . CRC press,2014. 1, 4 © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process [NB00] N
ADEAU
D. R., B
AILEY
M. J.: Visualizing volume data usingphysical models. In
Proceedings Visualization 2000. VIS 2000 (Cat. No.00CH37145) (2000), IEEE, pp. 497–500. 3, 5, 12, 14, 18, 19[NB15] N
ISSEN
B., B
OWERS
J.: Data-things: digital fabrication situ-ated within participatory data translation activities. In
Proceedings of the33rd Annual ACM Conference on Human Factors in Computing Systems (2015), Association for Computing Machinery (ACM), pp. 2467–2476.8, 20[NLC ∗
15] N
ITTALA
A. S., L I N., C
ARTWRIGHT
S., T
AKASHIMA
K.,S
HARLIN
E., S
OUSA
M. C.: Planwell: spatial user interface for col-laborative petroleum well-planning. In
SIGGRAPH Asia 2015 MobileGraphics and Interactive Applications . Association for Computing Ma-chinery (ACM), 2015, pp. 1–8. 4, 12, 15, 20[NTWVD19] N
ACHTIGALL
T., T
OMICO
O., W
AKKARY
R., V AN D ON - GEN
P.: Encoding materials and data for iterative personalization. In
Proceedings of the 2019 CHI Conference on Human Factors in Com-puting Systems (2019), Association for Computing Machinery (ACM),pp. 1–12. 7, 14, 20[OPSR16] O
MIROU
T., P
EREZ
A. M., S
UBRAMANIAN
S., R
OUDAUT
A.: Floating charts: Data plotting using free-floating acoustically levi-tated representations. In (2016), IEEE, pp. 187–190. 6, 16, 20[OS10] O
LSEN
L., S
AMAVATI
F. F.: Image-assisted modeling fromsketches. In
Proceedings of Graphics Interface 2010 (Toronto, Ontario,Canada, 2010), GI ’10, Canadian Information Processing Society / ACM,pp. 225–232. 10[OSSJ05] O
LSEN
L., S
AMAVATI
F. F., S
OUSA
M. C., J
ORGE
J. A.:Sketch-based mesh augmentation. In
Proceedings of the 2nd Euro-graphics Workshop on Sketch-Based Interfaces and Modeling (SBIM’05) (Dublin, Ireland, 2005), Eurographics Association, pp. 43–52. 10[OSSJ08] O
LSEN
L., S
AMAVATI
F. F., S
OUSA
M. C., J
ORGE
J.: A tax-onomy of modeling techniques using sketch-based interfaces. In
Euro-graphics 2008 - State of the Art Reports (STARs) (Crete, Greece, 2008),Eurographics Association, pp. 39–57. 10[OTS ∗
18] O H H., T A T. D., S
UZUKI
R., G
ROSS
M. D., K
AWAHARA
Y., Y AO L.: Pep (3D printed electronic papercrafts) an integrated ap-proach for 3D sculpting paper-based electronic devices. In
Proceedingsof the 2018 CHI Conference on Human Factors in Computing Systems (2018), Association for Computing Machinery (ACM), pp. 1–12. 14, 20[OW18] O
EHLBERG
L., W
ILLETT
W.: Encoding data through experien-tial material properties. In
Toward a Design Language for Data Physi-calization: Workshop at IEEE VIS 2018 (2018), IEEE. 21[Per14] P
EROVICH
L.: Data clothing: Dresses show air pollution,2014. Last Accessed Jan 2021. URL: http://dataphys.org/list/data-clothing-dresses-show-air-pollution/. 5, 6, 13, 19[PGDG12] P
RIESTNALL
G., G
ARDINER
J., D
URRANT
J., G
OULDING
J.: Projection augmented relief models (parm): Tangible displays for ge-ographic information.
Electronic Visualisation and the Arts (EVA 2012) (2012), 180–187. 3, 4, 7, 13, 14, 17, 19[PPW18] P
ANOTOPOULOU
A., P
ARIS
S., W
HITING
E.: Watercolorwoodblock printing with image analysis.
Computer Graphics Forum 37 ,2 (2018), 275–286. 8, 11, 13, 14, 20[PTPM17] P
ATEL
S. V., T AM K.-M. M., P
USHPARAJAN
S., M
IGNONE
P. J.: 3D sampling textures for creative design and manufacturing. In
Disciplines and Disruption - Proceedings Catalog of the 37th AnnualConference of the Association for Computer Aided Design in Architec-ture, ACADIA 2017 (2017), CUMINCAD. 4, 8, 13, 18, 20[Ras11] R
ASE
W.-D.: Creating physical 3D maps using rapid prototyp-ing techniques. In
True-3D in Cartography . Springer, 2011, pp. 119–134.4, 5, 7, 13, 19[RD14] R
EZAEIAN
A., D
ONOVAN
J.: Design of a tangible data visual-ization. In
Proceedings of the 7th International Symposium on Visual In-formation Communication and Interaction (2014), Association for Com-puting Machinery (ACM), pp. 232–235. 5, 19 [RG13] R
EYES -G ARCIA
E.: Motion structures: Videos as space-timeobjects, 2013. Last Accessed Jan 2021. URL: http://dataphys.org/list/motion-structures-videos-as-space-time-objects/. 8, 19[Rod18] R
ODIGHIERO
D.: Printing walkable visualizations. In (2018), Universityof Edinburgh, pp. 58–73. 8, 20[Row12] R
OWE
A.: Within an ocean of light: creating volumetriclightscapes. In
ACM SIGGRAPH 2012 Art Gallery (2012), Associationfor Computing Machinery (ACM), pp. 358–365. 8, 19[RPPH12] R
ASMUSSEN
M. K., P
EDERSEN
E. W., P
ETERSEN
M. G.,H
ORNBÆK
K.: Shape-changing interfaces: A review of the designspace and open research questions. In
Proceedings of the SIGCHIConference on Human Factors in Computing Systems (New York,NY, USA, 2012), CHI ’12, Association for Computing Machinery,p. 735–744. URL: https://doi.org/10.1145/2207676.2207781, doi:10.1145/2207676.2207781 . 15[Rüs14] R
ÜST
A.: A piece of the pie chart: feminist robotics. In
ACMSIGGRAPH 2014 Art Gallery . Association for Computing Machinery(ACM), 2014, pp. 360–366. 3, 6, 16, 19[Sam13] S
AMAVATI
F.: Modelling for computer graphics - cpsc 589/689course notes., 2013. 10[SB16] S
TOPPEL
S., B
RUCKNER
S.: Vol velle: Printable interactive vol-ume visualization. IEEE transactions on visualization and computergraphics 23 , 1 (2016), 861–870. 11, 18, 20[Sch04] S
CHMITZ
M.: XenoVision Mark III: A dynamic solid terrainmodel, 2004. Last Accessed Jan 2021. URL: http://dataphys.org/list/xenovision-mark-iii-a-dynamic-solid-terrain-model/. 7, 15, 19[Sch15] S
CHNEIDER
D.: Visualisation de projets de thèse avec undispositif physique et constructif. In (Agadir, Morocco, June 2015), pp. 447–449. Conférence EIAH 2015.URL: https://hal.archives-ouvertes.fr/hal-01405987. 5, 8[Sch16] S
CHWARTZ
C.:
Introducing architectural tectonics: Exploringthe Intersection of Design and Construction . Taylor & Francis, 2016. 9[Seg11] S
EGAL
A.: Adrien Segal’s data furniture, 2011. Last AccessedDec 2020. URL: http://dataphys.org/list/data-furniture/. 5, 6, 7, 9, 17,19[Seg12] S
EGERMAN
H.: 3D printing for mathematical visualisation.
TheMathematical Intelligencer 34 , 4 (2012), 56–62. 1[SGE ∗
18] S
OSA
R., G
ERRARD
V., E
SPARZA
A., T
ORRES
R., N
APPER
R.,
ET AL .: Data objects: Design principles for data physicalisation.In
DS 92: Proceedings of the DESIGN 2018 15th International DesignConference (2018), pp. 1685–1696. 9[SJK ∗
17] S
WACKHAMER
M., J
OHNSON
A. J., K
EEFE
D., J
OHNSON
S.,A
LTHEIMER
R., W
ITTKAMPER
A.: Weather report: Structuring dataexperience in the built environment. In
Proceedings of Architectural Re-search Centers Consortium (2017), pp. 102–111. 8[ŠLH ∗
14] Š
IMBELIS
V., L
UNDSTRÖM
A., H
ÖÖK
K., S
OLSONA
J.,L
EWANDOWSKI
V.: Metaphone: machine aesthetics meets interactiondesign. In
Proceedings of the SIGCHI Conference on Human Factorsin Computing Systems (2014), Association for Computing Machinery(ACM), pp. 1–10. 4, 5, 19[Som14] S
OMETHING &S ON : People wood: Data sculpture of question-naire data, 2014. Last Accessed Dec 2020. URL: http://dataphys.org/list/people-wood-data-sculpture-of-questionnaire-data/. 8, 12, 19[Son21] S ONDERMANN
H.: Parametric bim school, 2021.Last Accessed Feb 2021. URL: https://horstsondermann.com/rhino3d-grasshopper-dispatch-list-match-text/. 10[SPG ∗
16] S
CHÜLLER
C., P
ANOZZO
D., G
RUNDHÖFER
A., Z
IMMER
H., S
ORKINE
E., S
ORKINE -H ORNUNG
O.: Computational thermo-forming.
ACM Transactions on Graphics (TOG) 35 , 4 (2016), 43. 8,11, 13, 14, 20 © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process [SRK ∗
16] S
WAMINATHAN
S., R
OUMEN
T., K
OVACS
R., S
TANGL
D.,M
UELLER
S., B
AUDISCH
P.: Linespace: A sensemaking platform forthe blind. In
Proceedings of the 2016 CHI Conference on Human Fac-tors in Computing Systems (2016), Association for Computing Machin-ery (ACM), pp. 2175–2185. 4, 8, 20[SRP18] S
TARRETT
C., R
EISER
S., P
ACIO
T.: Data materialization: ahybrid process of crafting a teapot.
Leonardo 51 , 4 (2018), 381–385. 6,8, 12, 20[SSB15] S
TUSAK
S., S
CHWARZ
J., B
UTZ
A.: Evaluating the memora-bility of physical visualizations. In
Proceedings of the 33rd Annual ACMConference on Human Factors in Computing Systems (2015), Associa-tion for Computing Machinery (ACM), pp. 3247–3250. 1[SSJ ∗
14] S
WAMINATHAN
S., S HI C., J
ANSEN
Y., D
RAGICEVIC
P.,O
EHLBERG
L. A., F
EKETE
J.-D.: Supporting the design and fabricationof physical visualizations. In
Proceedings of the SIGCHI Conference onHuman Factors in Computing Systems (2014), ACM, pp. 3845–3854. 3,4, 8, 11, 13, 16, 17, 20, 21[Str14] S
TREAMERS
D. D.: Drip-by-tweet: Each vote is a drop,2014. Last Accessed Jan 2021. URL: http://dataphys.org/list/drip-by-tweet-each-vote-is-a-drop/. 6, 7, 19[STS ∗
14] S
TUSAK
S., T
ABARD
A., S
AUKA
F., K
HOT
R. A., B
UTZ
A.: Activity sculptures: Exploring the impact of physical visualizationson running activity.
IEEE Transactions on Visualization and ComputerGraphics 20 , 12 (2014), 2201–2210. 3, 5, 6, 12, 19[Stu08] S
TUDIOS
A.: Bmw kinetic sculpture, 2008. Last Accessed Jan2021. URL: http://dataphys.org/list/bmw-kinetic-sculpture/. 8, 16, 17,19[Stu13a] S
TUSAK
S.: Layered 2d plots, 2013. Last Accessed Jan 2021.URL: http://dataphys.org/list/layered-2d-plots/. 13, 17[Stu13b] S
TUSAK
RENT
S., A
NDREW
N., C
HELSEA
J., S
TEVEN
P.: Fizviz:A wall-mounted data gauge, 2016. Last Accessed Jan 2021. URL: http://dataphys.org/list/fizviz-a-wall-mounted-data-gauge/. 8, 9, 10, 14, 20[TGZ18] T
YMMS
C., G
ARDNER
E. P., Z
ORIN
D.: A quantitative per-ceptual model for tactile roughness.
ACM Transactions on Graphics(TOG) 37 , 5 (2018), 1–14. 4, 8, 18, 20[THHC18] T
HUDT
A., H
INRICHS
U., H
URON
S., C
ARPENDALE
S.:Self-reflection and personal physicalization construction. In
Proceedingsof the 2018 CHI Conference on Human Factors in Computing Systems (2018), Association for Computing Machinery (ACM), pp. 1–13. 21[THK ∗
15] T
AHER
F., H
ARDY
J., K
ARNIK
A., W
EICHEL
C., J
ANSEN
Y., H
ORNBÆK
K., A
LEXANDER
J.: Exploring interactions with phys-ically dynamic bar charts. In
Proceedings of the 33rd Annual ACMConference on Human Factors in Computing Systems (2015), ACM,pp. 3237–3246. 4, 6, 7, 15, 20[THP ∗
17] T
ABRIZIAN
P., H
ARMON
B., P
ETRASOVA
A., P
ETRAS
V.,M
ITASOVA
H., M
EENTEMEYER
R.: Tangible immersion for ecologicaldesign. In “Disciplines and Disruption" Proceedings of the 37th AnnualConference of the Association for Computer Aided Design in Architec-ture (ACADIA) (2017), CUMINCAD, pp. 600–609. 13[TJW ∗
16] T
AHER
F., J
ANSEN
Y., W
OODRUFF
J., H
ARDY
J., H
ORN - BÆK
K., A
LEXANDER
J.: Investigating the use of a dynamic physicalbar chart for data exploration and presentation.
IEEE transactions onvisualization and computer graphics 23 , 1 (2016), 451–460. 1, 6, 7, 20[TL16] T
HRUN
M. C., L
ERCH
F.: Visualization and 3D printing of mul-tivariate data of biomarkers. In
The 24th International Conference inCentral Europe on Computer Graphics, Visualization and Computer Vi-sion 2016 in co-operation with EUROGRAPHICS: University of WestBohemia, Plzen, Czech Republic (2016), Václav Skala-UNION Agency,pp. 7–16. 4, 5, 12, 20[TMB ∗
13] T
ENA
J. R., M
AHLER
M., B
EELER
T., G
ROSSE
M., Y EH H., M
ATTHEWS
I.: Fabricating 3D figurines with personalized faces.
IEEE computer graphics and applications 33 , 6 (2013), 36–46. 8, 19 [TMH ∗
10] T
ATEOSIAN
L., M
ITASOVA
H., H
ARMON
B., F
OGLEMAN
B., W
EAVER
K., H
ARMON
R.: Tangeoms: Tangible geospatial model-ing system.
IEEE transactions on visualization and computer graphics16 , 6 (2010), 1605–1612. 4, 7, 14, 17, 19[TSW ∗
19] T
ANG
K., S
ONG
P., W
ANG
X., D
ENG
B., F U C.-W., L IU L.: Computational design of steady 3D dissection puzzles. In
ComputerGraphics Forum (2019), vol. 38, Wiley Online Library, pp. 291–303. 8,20[TVR ∗
12] T
HÜN
G., V
ELIKOV
K., R
IPLEY
C., S
AUVÉ
L., M C G EE W.:Soundspheres: resonant chamber.
Leonardo 45 , 4 (2012), 348–357. 8,17, 19[TWD ∗
18] T
HOMAS
B., W
ELCH
G., D
RAGICEVIC
P., E
LMQVIST
N.,I
RANI
P., J
ANSEN
Y., S
CHMALSTIEG
D., T
ABARD
A., E L S AYED
N.,S
MITH
R., W
ILLETT
W.: Situated Analytics. In
Immersive Analytics ,vol. 11190 of
Lecture Notes in Computer Science . Springer, Oct. 2018,pp. 185–220. URL: https://hal.inria.fr/hal-01947243, doi:10.1007/978-3-030-01388-2\_7 . 15[VKBR ∗
18] V AN K OLLENBURG
J., B
OGERS
S., R
UTJES
H., D
ECK - ERS
E., F
RENS
J., H
UMMELS
C.: Exploring the value of parent trackedbaby data in interactions with healthcare professionals: A data-enableddesign exploration. In
Proceedings of the 2018 CHI Conference on Hu-man Factors in Computing Systems (2018), Association for ComputingMachinery (ACM), pp. 1–12. 17, 20[VM08] V
ANDE M OERE
A.: Beyond the tyranny of the pixel: Exploringthe physicality of information visualization. In (2008), IEEE, pp. 469–474. 1[VTOS14] V
ELIKOV
K., T
HÜN
G., O’M
ALLEY
M., S
IMBUERGER
W.:Nervous ether: soft aggregates, interactive skins.
Leonardo 47 , 4 (2014),344–351. 8, 10, 13, 17, 19[Whi09] W
HITELAW
M.: Mitchell whitelaw’s weather sculptures,2009. Last Accessed Jan 2021. URL: http://dataphys.org/list/weather-bracelet-and-measuring-cup/. 8, 12, 19[WOSC19] W UN T., O
EHLBERG
L., S
TURDEE
M., C
ARPENDALE
S.:You say potato, i say po-data: Physical template tools for authoring visu-alizations. In
Proceedings of the Thirteenth International Conference onTangible, Embedded, and Embodied Interaction (2019), Association forComputing Machinery (ACM), pp. 297–306. 21[WPHC16] W UN T., P
AYNE
J., H
URON
S., C
ARPENDALE
S.: Compar-ing bar chart authoring with microsoft excel and tangible tiles. 111–120.1, 21[WWOC19] W
ANNAMAKER
K., W
ILLETT
W. J., O
EHLBERG
L. A.,C
ARPENDALE
S.: Data embroidery: Exploring alternative mediums forpersonal physicalization, 2019. URL: https://prism.ucalgary.ca/handle/1880/110218. 21[YOC ∗
15] Y AO L., O U J., C
HENG
C.-Y., S
TEINER
H., W
ANG
W.,W
ANG
G., I
SHII
H.: Biologic: natto cells as nanoactuators for shapechanging interfaces. In
Proceedings of the 33rd Annual ACM Confer-ence on Human Factors in Computing Systems (2015), Association forComputing Machinery (ACM), pp. 1–10. 8, 14, 17, 20[ZC18] Z
ORAN
A., C
OHEN
D.: Digital konditorei: programmable tastestructures using a modular mold. In
Proceedings of the 2018 CHI Con-ference on Human Factors in Computing Systems (2018), Associationfor Computing Machinery (ACM), pp. 1–9. 8, 12, 13, 20[ZDS16] Z
HANG
L., D
ONG
H., S
ADDIK
A. E.: From 3D sensing toprinting: A survey.
ACM Transactions on Multimedia Computing, Com-munications, and Applications (TOMM) 12 , 2 (2016), 27. 12[ZDX ∗
18] Z
HANG
X., D
EKEL
T., X UE T., O
WENS
A., H E Q., W U J.,M
UELLER
S., F
REEMAN
W. T.: Mosculp: Interactive visualization ofshape and time. In
Proceedings of the 31st Annual ACM Symposium onUser Interface Software and Technology (2018), Association for Com-puting Machinery (ACM), pp. 275–285. 8, 12, 18, 20[ZLW ∗
16] Z
HAO
H., L U L., W EI Y., L
ISCHINSKI
D., S
HARF
A.,C
OHEN -O R D., C
HEN
B.: Printed perforated lampshades for contin-uous projective images.
ACM Transactions on Graphics (TOG) 35 , 5(2016), 1–11. 8, 12, 20 © 2021 The Author(s)Computer Graphics Forum © 2021 The Eurographics Association and John Wiley & Sons Ltd. javaherpour et al. / Data to Physicalization: A Survey of the Physical Rendering Process [ZYZZ15] Z
HANG
Y., Y IN C., Z
HENG
C., Z
HOU
K.: Computationalhydrographic printing.
ACM Transactions on Graphics (TOG) 34 , 4(2015), 1–11. 8, 11, 20, 4(2015), 1–11. 8, 11, 20