 Good afternoon. We are Nathan Hall of Virginia Tech and Jamie Wittenberg of Indiana University. We're representing the IMLS funded library strategies for 3D and virtual reality collection development and reuse project. For more information we have our grant information at the bottom of the page and our contact information at the end of the slides. And we also want to mention another project, the community standards for 3D preservation and reuse, another IMLS funded project that is working separately but in a complementary and collaborative way with us. So I thought I'd start with a rationale for the project. Why should libraries be involved in 3D or virtual reality technology? So for starters, advances in 3D technologies have expanded possibilities in a number of fields disciplines. At the same time costs are lower so it's accessible to more people and so it results in a wild west situation. So for our overview and goals, as a result of that of those reasons, crosswalks and access between standards align with the strengths of libraries while supporting research and education lines with the mission of libraries. And so therefore the goals of our project is to form holistic knowledge from normally silent areas of 3D and virtual reality, develop best practices to support 3D virtual reality throughout the research life cycle, establish guidelines that can serve multiple research contexts and use cases that libraries may need to support as researchers increasingly adopt 3D and VR as a research tool, and develop strategies that libraries can use for creating policies and workflows. So this is an inter-institutional project with members from Virginia Tech, University of Oklahoma, and Indiana University. And then we have an advisory board representing, it's a diverse advisory board representing multiple sectors and disciplines and they've advised us on our goals, our research questions, and they've expanded our network for possible participants and project outcomes. This slide represents the diversity of our partners from a number of sectors and disciplines. So green for the Washington D.C. form that we held, blue for the Chicago form that we held, and yellow for the Oklahoma form. And then this slide represents the disciplinary diversity of the participants across the forums and its representation of only the researchers who participated in the forums themselves. And I'm not going to read through these or spend much time on these slides, but for future reference, if you're interested, we have the list of our D.C.s across the three forums, Washington D.C., Oklahoma, and Chicago. For a high-level overview of each forum, for Washington D.C., we focused on 3D and virtual reality content creation and publishing. For Oklahoma, we focused on 3D and virtual reality visualization and analysis. And for the Chicago forum, we focused on repository standards and practices. From the D.C. forum, we identified tensions between 3D and virtual reality data for use in research education exhibition versus that for long-term access and preservation. So for an accurate representation for scientific inquiry, it requires more granular data, more accurate, transparent capture, and processing. And these all translate to more labor and more sophisticated tools throughout the image capture, processing, and documentation processes. So from this image of a beetle next to a diamond to give you scale, here's a series of photos. Well, this is a screenshot from the 3D imaging software, Agisoft. On the left pane, you'll see a series of photos of beetles taken from different angles. And then each angle is represented by a blue polygon in the center frame. And so you see there's a kind of a little cloud in the middle. This is zoomed way out, so you can see the different camera angles. And each of these beetle photos, the software finds matching points across them and creates these points in a point cloud. And as I said, we're zoomed really far out, zoomed really far into the model later on. As I said, greater transparency for research integrity and reproducibility requires larger datasets and more labor, particularly in the texture art side. So in this image, we see some technical artifacts from the processing methods. That top horn on this beetle should have hairs on it. But the software mistakenly interpreted that texture as what appears to be a somewhat waxy crystalline substance. Now a preservationist would argue for keeping that. That's an artifact of the process itself. A texture artist could clean it up and make it look right. And a texture artist, by the way, is what they call these folks. It's typically in the gaming industry, so people who kind of polish 3D models. And so a texture artist could clean that up and make it look right. A scientist would not want the data altered without documenting the changes, but an educator might want a cleaner looking model. So with all these different needs, you have an archival challenge. You have to decide what to keep and for whom. And photogrammetry offers better transparency than laser scanning because there are fewer proprietary technologies involved. But it makes for more expensive labor, so you can't produce as many models in the same amount of time. And because each one requires taking hundreds of photos and then time to process them and then the storage that goes along with it. So basically how is 3D data managed? So at its most basic, 3D data can be stored in a spreadsheet with XYZ values to represent height, width, and depth. So here's basically a graph of three vectors. And then for color, it's using familiar ways as well. This is an RGB standard. So there are three channels of color, red, green, and blue, each with their own values from 0 to 255. That together create nearly 17 million possible color values, 0 in all channels and all three channels will be black, 255 in all three channels is white, and other values produce various shades and intensities of color. So put together, you can represent color and space with XYZ and RGB. So each XYZ point is aligned with RGB color, and then you have multiple points in a volume of space, and then you have a point cloud. And so you can connect to the dots then to map the surface as an interconnected set of polygons that represents the overall shape and structure of whatever it is you've created. And this by itself is fairly straightforward from a data management perspective, even if you have millions of these data points because computers are good at storing and managing this kind of information. But if you want to interpret or make knowledge claims about what that data means or why aspects of it are significant, you need expert reviewers who understand the data to evaluate those claims using standardized methods. And we're still missing the infrastructure to be able to do that. I'm going to pass to Jamie now who's going to discuss some of the infrastructure challenges. Thanks, Nathan. Like Nathan said, I'm Jamie Wittenberg from Indiana University, and I'm going to zoom in a little bit for you on the third forum that this project posted in Chicago. So as a reminder, that project focused on 3D and VR repositories and repository infrastructure. And essentially the project team has been working as part of that forum and sense on finding answers to three primary questions that I put up there on the screen for you. So what features are really necessary in a repository to support 3D data management? What workflows or best or better practices can be applied to 3D data to support that storage and management? And what are the implications of 3D data and archiving or collecting 3D data on current preservation models? So the models that we're already using in our repositories. So to achieve this, 16 participants were invited to the forum, and they had affiliations from across academia, libraries, museums, the repository software industry, and government. And you saw those breakdowns of participants that Nathan showed you earlier in our slides. So there was a lot of diversity at these forums. So the first thing that we wanted to do is ensure that everyone was kind of on the same page. And what we did is we looked at case studies and workflow diagrams from existing 3D VR repositories. Successful repositories already supporting 3D and VR data. So these included York University's Archaeology Data Service, Duke University's Morpho Source, which is a paleontology repository, University of Oklahoma's 3D repository, and the NIH 3D print exchange. So all of these repositories have sort of very different user communities. And that was really helpful to the participants who then went on to prioritize repository features that they found were essential in supporting 3D and VR. And in general, these features kind of fell into broader categories of repository sort of essential features that included preservation, searchability or discoverability, download capacity, streaming capability, security, and workflows. So once those repository features were kind of prioritized and the workflow diagrams were reviewed, the participants went on to discuss implications for different levels of library support and the implications for existing preservation models and practices. And what we did is we mapped those features or those requirements to some really kind of widely adopted models for digital preservation curation and repository support. So they reviewed three models for 3D VR presentation and many of you will be familiar with these models. I'll only give sort of brief summaries for those of you who aren't. So the first was the DCC curation life cycle model. So the DCC life cycle model, it represents a set of sort of actions and tasks, some of which are sequential, that are necessary to effectively curate and preserve digital content from what they call conceptualization to disposition. So full life cycle actions, which are description and representation, information, preservation planning, community watch and participation, and curate and preserve, along with the model's occasional actions, dispose, reappraise, and migrate, were largely accepted by the group as useful and necessary for the preservation of 3D and VR. Regarding the sequential actions, and those are the actions that are in red that you see in the center or in the sort of outer edge of the circle. Those include conceptualize, create or receive, appraise and select, ingest, preservation action, store, access, use, and reuse, and transform. Participants express that the unique nature of 3D and VR content might require some modification of this model to successfully support this kind of data. So one of the things that they noted was that 3D models, the way that 3D models are generated often uses really proprietary systems. So commercial software and transforming data repeatedly often does not produce the same results when you're talking about 3D. So in the case of VR, participants share that it's really difficult to decouple the digital content from the hardware requirements and the necessity of capturing hardware and environment isn't really adequately represented in this model. So they felt that applying 3D kind of data life cycles to this model might create some blockages in the sequential paths that are shown here, especially due to the lack of transparency and openness associated with file processing. So there was some discussion around how useful the DCC curation lifecycle model might be for 3D and VR and repositories. The second model that the group reviewed is the OAIS reference model. So this conceptual framework defines three types of information packages that move through the system, the submission information package or SIP, archival information package or AAP, and dissemination information package or DIP. So in the case of 3D, VR, these packages can be mapped to systems as well as users or stakeholders, right? So where the creator of the object, the creator of the 3D model submits the SIP, the data manager or repository manager kind of assembles the AAP and the reader or patron accesses the DIP. So assessing the usefulness of this model gave rise to a lot of concern about limitations in the framework, especially for dissemination. Particularly, there's a difference between experiencing the 3D VR medium through playback and downloading a package for reuse specifically. And these are both types of access, but they really have different requirements for access and viewing and for downloading and reusing. And they can be very different types of digital objects depending on how they're accessed. And they can also be from completely different storage sources. So for 3D VR, this is kind of a more nuanced view of access for the different ways that it might be delivered. And there was some skepticism about how useful, especially the dissemination package model, might be in this reference model. Finally, participants reviewed the fair data principles published in 2016. So there are a series of kind of guiding principles that aim to improve the findability, accessibility, interoperability, and reproducibility of data. And these principles were probably the most controversial of all the models that we discussed. Participants really reiterated that interoperability is a big challenge. And research around fair data principles has also shown that interoperability and reuse or reusability is really difficult to assess and implement in repositories and data repositories. So that's not too surprising. But the big challenge with interoperability is, again, due to the proprietary technologies of developing 3D and VR models. So participants really questioned whether a metric requiring that data and metadata use formal, accessible, shared, and broadly applicable language for knowledge representation is achievable for 3D VR. So they concluded that it might be achievable for metadata, but likely is not for data where some details are really contested, like the term mesh is a good example. So individuals and communities really disagree on what the formally defined terms are for 3D VR, like medical CT and industrial CT, for example, have different vocabularies. So participants suggested that this could be maybe an area for better rather than best practices. And they also discussed potentially the extension of FAIR principles to FAIRST, F-A-I-R-E-S-T, in order to really represent the importance of ethical, sustainable, and transparent practices in 3D and VR. They also discussed the necessity of representing the importance of an object being playable or instantiable as a discrete metric in this model. So one of the most important features that was discussed across these models is capturing intellectual property for 3D and VR. There's really a significant amount of confusion and misinformation related to copyright and IP in the context of 3D VR. So some experts agree that 3D scanning might not result in copyrightable IP, especially if it's a sort of slavish copy, which is the term characterizing non-transformative scans in the 1999 Bridgeman Art Library versus Curl Corporation case. And for more on that, Michael Weinberg wrote a 2017 white paper entitled 3D Scanning a World Without Copyright. That's a great reference for that perspective. Kyle Courtney and Melissa Levine also presented on this issue at the Community Standards for 3D Preservation Meeting in St. Louis. That's the kind of compatible 3D VR project that is also funded by IMLS that Nathan mentioned at the beginning of this presentation. And that was in early 2018. And they really highlighted the complexity of the question and since have been working on a white paper to better detail the issue. But lib3D VR participants throughout the three forums really expressed uncertainty about the IP status of 3D models and replica and printed objects and the implications that uncertainty might have for organizations working to archive and disseminate those models. In particular, the issue of models derived from hundreds of crowdsourced images, so photogrammetric models that are using all of these images. Yeah. All with different licenses really pose a problem. So here's an example. This is a building room on a cloudless day. So in 2010, the University of North Carolina in partnership with ETH Zurich developed technology that allowed them to stitch together millions of flicker images to create the 3D models that you see here. So models of landmarks and cities. So the project and accompanying paper, Building Room on a Cloudless Day, uses the city of Rome as a kind of prototype for this sort of crowdsourced modeling. And the authors wrote, our method efficiently combines 2D appearance and color constraints with 3D multi-view geometry constraints to estimate the geometric relationship between millions of images. And you can see here they're using almost 3 million image collections in order to stitch together these models. And they're using flicker as source data for them. So there could be any number of licenses that are applied to these images. And this is really the kind of scale that we're talking about when we express concern about IP and copyright in particular as it relates to 3D and VR. I'm going to pass it back to Nathan to discuss other projects that have implications for copyright and ethics. So Photosynth created photogrammetric data sets from images created from flicker. As Jamie mentioned, here's another example from an inauguration to that. And I think this is maybe the 2008. But Microsoft started doing this with Photosynth back in 2005. So yeah, as Jamie mentioned, there's copyright issues there. And then aside from copyright, there are times where it's totally legal but ethically questionable. So a 3D image capture of a public cultural heritage site is legal, but commercializing it or making open access creates ethical challenges. So in this example, a company called CyArc created a virtual environment of a temple complex in Myanmar. Does CyArc own it? CyArc says it copyrighted the scans so that no one would use them in an inappropriate way. So that's good because you don't want people to take a set like that and create a first person shooter video game there. But meanwhile, Google provides access to it for free through arts and culture, but they're commercializing it through visitor stats and advertising. But what about Myanmar or any other cultural heritage site? How do they benefit from this kind of data capture? Some people would counter that Myanmar currently has a government which is arguably not a good negotiating partner to represent their people. But what else is there? Does access to this site raise awareness among the public? But who benefits and how? So this is related to a term referred to as digital colonialism, which Renata Avila, I believe, either coined or described as a quasi-imperial power over a vast number of people without their explicit consent, manifested in rules, designs, languages, cultures, and belief systems by a vastly dominant power. So on the other end, you have digital repatriation, which is return of images of cultural heritage in a digital format to the communities from which they originated. So participants like the idea of platforms like Mukurtu, particularly for built environment and cultural heritage objects, structures and landscapes associated with indigenous groups. So the Mukurtu project addresses some of these issues by letting communities set user classes and groups along with who should have access to which artifacts. So you could have certain objects that are only accessible by, say, female elders of a particular community or maybe Shaman, or maybe it's only accessible at certain times of the year to be when it's appropriate for a certain season but not appropriate for others. So, but digital repatriation doesn't fix everything and it introduces new questions and new problems, but most importantly, it encourages working with the communities that the artifacts come from. So sort of our summative explanations of some of the issues we've discussed and explored. There are a number of implications for repository systems developed by academic libraries in the context of 3D and virtual reality data, and these relate to metadata, to existing infrastructure and how we adapt it, privacy, ethics, intellectual property, and use for research teaching and exhibition. And from there we hear our names and contact information again and we'd be happy to spend the last few minutes taking your questions. Thank you very much.