Europeana’s strongest asset is the Network which supports and contributes to it. The collaboration and cooperation between the Network, the transfer of data, adherence to policies and continued progress has allowed Europeana to come to fruition and continue to change the world with cultural heritage. Over the past few years Europeana has combined efforts with similar international efforts like DPLA, DigitalNZ and Trove but few of these combined efforts have directly related to technical infrastructure development. That’s why EuropeanaTech is so excited about joining the International Image Interoperability Framework (IIIF), a technical framework that in the past several years has propelled the interoperability and presentation of high quality images to a new level of collaboration.

More recently, EuropeanaTech has launched a Task Force to assess the needs and requirements of the Tech community to propel a community driven movement towards universal adoption and implementation of IIIF and its subsequent open source software. In the meantime, with this special issue of EuropeanaTech Insight, we invited several institutions familiar with IIIF to share their experiences discovering, adopting and deploying IIIF, expressing both struggles, of which there are few, and successes, of which there are many!

We hope that hearing from your fellow EuropeanaTech community members will entice you and inspire you and your institute to investigate IIIF further so that the Europeana Network’s image sharing and presentation can grow to a singular, interoperable web, benefitting every institute, their users and anyone interested in celebrating heritage imagery.

The Library at University College Dublin has, for more than a decade, been creating digital content that is disseminated via UCD Digital Library. Since the majority of digital file assets consist of images—high-resolution TIFF images with lossless JPEG2000 and lower-resolution static JPEG derivatives—the emergence of the International Image Interoperability Framework immediately attracted our attention. It offered a common-sense approach to image interoperability that proposes to address technical and usability challenges that inhibit both service development and scholarly interaction with digital content.

UCD faced challenges typical with providers of digitised books, maps, and other cultural heritage information: use of older technologies for dissemination of deep-zooming images (djatoka, whose OpenURL-based approach to interoperability lost lustre with the domination of approaches based on REST); challenges in delivering paged documents, including page turning and document navigation functionality, as well as associating OCR content with page images; a large-scale transcription project based on bespoke technologies; and more. IIIF and its growing community of developers promised solutions for all of these issues, and as they have been delivered over the past five years, we’ve adopted them: IIIF Image API (Loris), IIIF Presentation, and IIIF Search; IIIF authentication is next on the development roadmap.

Figure 1. IIIF search via UniversalViewer

Exploiting the potential for interoperability offered by IIIF is where we have set our sights for current and future development. In 2016, we seized an opportunity to work with Nuno Freire and Pierre Edouard-Barrault at Europeana to work on a case study of reporting our IIIF content to Europeana. The project was experimental (also involving the National Library of Wales), evaluating approaches and testing ideas. But it also has succeeded in a practical sense, delivering new content from UCD and its partner organisations in Ireland to Europeana. It has catalysed other improvements as well—including creation of a microservice to deliver EDM metadata via IIIF manifests, and the upgrading of local sitemaps to use components of the ResourceSync framework. (We’ll next be testing with Europeana an approach to reporting video content via sitemaps and the EDM microservice.)

Figure 2. UCD content in Europeana via IIIF

Extending interoperability more aggressively to end-users is UCD Library’s next major objective. UCD Library supports a major document transcription project in collaboration with Fiontar at Dublin City University, hosted at http://duchas.ie. Underway for several years, it uses bespoke technologies for engaging end-users in document transcription, and has been extraordinarily successful in the degree of public engagement it has fostered.

Figure 3. Screenshot from duchas.ie transcription site

As we look towards the future, we see IIIF as a foundational technology in such projects exposing our content to scholarly communities and the public at large. The integrated annotation capabilities of the Mirador image viewer will be invaluable in enabling annotation of more than 70,000 images from the Irish National Folklore Collection, due to come online in 2017. And as IIIF client capability becomes increasingly integrated with tools such as Scribe, FromThePage, or Europeana’s transcription tools, we see vast potential for additional transcription projects that focus on structured textual documents, whether printed or manuscript.

IIIF does not come without its challenges. For library staff accustomed to working with the ubiquitous standards of the Library community, there is a learning curve that is partly specific to IIIF, but which more generally has to do with re-orienting one’s thinking to web standards, rather than library standards for interoperability. Data representation with JSON/JSON-LD is one dimension of this; understanding Web Annotations is another, which is crucial to understanding the general direction of web-based document transcription and annotation; and there are the complexities of web messaging and notifications that the IIIF community continues to grapple with.

The challenges also extend to how new technologies that are driven by IIIF and web standards integrate with infrastructures based on older library practices. In UCD’s case, we digitise printed documents and capture OCR’d text in METS/ALTO format, indexed by Solr. Integrating the technologies supporting this approach with IIIF Search was a straightforward exercise. But what are the implications of deploying an or Web Annotations-compliant annotation store for data management and information retrieval, and especially for maintaining a coherent user experience? How do we make the old and the new work together harmoniously and sensibly?

Figure 4. Imagining an evolved systems architecture at UCD Digital Library

Five years ago, when we became aware of the concept of the International Image Interoperability Framework through Tom Cramer’s presentation at the Coalition for Networked information, we knew it had potential to be transformative. But as we look at our own work during the past two years and the development agenda for the near term, we at UCD can say that we simply did not anticipate the degree to which it would shape our overall agenda. As our technical infrastructure develops, we see it opening new doors not only to interoperability, but also to broader engagement with UCD’s academic community, its heritage repositories, our growing range of partners in the Irish cultural heritage community, and the public at large.

Although the IIIF-standard has been around for some time, it took us until April 2016 to realise the full potential it offers cultural heritage institutions and end-users. In 2016 during the museum’s 150-year-anniversary, Nationalmuseum was facing the problem that we wanted to provide access to some of our high resolutions images but did not have the possibility to largely invest into infrastructure. During a presentation at Museums and the Web we realised that IIIF was the most logical option as it would be the fastest to implement and by far the most future-proof solution to our problem. The IIIF image API in combination with a IIIF image server makes it possible to deliver, embed and request images breathtakingly fast in any desired size simply by embedding the image´s URI into the desired website.

Figure 1: Example of Nationalmuseum content with IIIF zoom

In the long run, this means that we will no longer need to keep derivatives of any images accessible on a IIIF server but would just embed the same URI with different size specifications into different websites, blogs, Social Media applications etc. There is no more need to download and upload images into different CMSs for different use cases, and no need for endless copies in different systems.

If we needed a small derivative of this image we would simply alter the URI, with no need to host or upload a different image somewhere else. It is even possible to cropor turn the image if necessary for any desired presentation. These possibilities are a great chance for us when planning the relaunch of our general website as well as our collections online presentation in 2017.

More so, with a suite of open source viewers accessible, a lot of additional functionalities come along with having images on a IIIF server without having to understand any of the underlying features or technologies at all. With our primary aim to “have a set of high-resolution images accessible in a user-friendly way in 2016”, we were facing two simple tasks: uploading the images in question to a IIIF hosting service and implementing the relevant URIs to the images as well as a IIIF image viewer into our online database. As there were just 3,000 images, we uploaded them manually and fed the URIs back into the database by a simple import script. The supplier of our online collection implemented OpenSeaDragon at a reasonable cost into the existing website without changing any of the existing design or functionality. Thus, whenever a IIIF URI is available for the object in question, additional functionalities will appear, and when there is no IIIF URI, the service will just look as it used to. or example, compare the portrait of Karen Margrethe Borch with the Portrait of a lady with a dog. Those images with a IIIF URI now offer deep zoom and full screen view for our users, who “love IIIF without knowing what it is”.

Figure 2: example of OpenSeaDragon image reference strip

We did not upload all our images to the IIIF hosting service due to the fact that we did not acquire enough hosting space from the beginning. Therefore, we are eager to try Europeana’s IIIF hosting server when it comes up. As we have our images in a cloud based DAMS, we do not use the IIIF server as an additional backup, and only uploaded jpgs to the hosting service, excluding tiffs. Nevertheless, we are planning to upload at least one image from any collection object to a IIIF server.

At the moment our usage of IIIF is reduced to the OpenSeadragon implementation in the online database, but we will explore all possibilities for both embedding IIIF into the new website, Social Media portals and online catalogues when relaunching the website and the online database. Another step will be to try to make use of the presentation API so that we can deliver metadata along with the images. Additionally the image presentation capabilities are much richer for end-users.

The presentation API delivers the necessary information to understand what the image represents, which means that it makes it possible to show relations between different images, making it easy to compare images (e.g. X-ray or infrared images of the same object) or to group images that belong together.

Recently there has been a new development that allows us to replace the edm:IsShownBy links for our objects in Europeana that have a IIIF URI available. The working group behind the MINT ingestion service worked on a transformation of the LIDO to EDM mapping which allows the delivery of IIIF links as part of our metadata delivery process. Doing so, the hosted image with all zoom functionalities will even be available from Europeana’s portal and will let users explore all astonishing details of paintings in the Nationalmuseum’s collection.

Over the last 3 years IIIF has been working to establish its position as the go-to solution for online Image content delivery within the Libraries, Archives and Museums sector, achieving remarkable results. IIIF accomplished standardising the performance optimisation of tiling image servers, thus improving the performance of accessing large high resolution images. IIIF is currently supported by the Image API (serving image content), the Presentation API (presenting the resource to the human reader), the Search API (enabling search over the resource) and the Authentication API (enabling the possibility of access restrictions). An extension to this successful track record is the development of a new framework that will be able to serve Audio and Video content, the IIIF A/V API.

IIIF A/V API has been in the make since April 2016, when the first A/V Working Group meeting took place at the British Library. The result was an extensive functionalities “wishlist”, crafted by domain experts who came up with numerous use cases that the new framework is expected to cover. In the meetings that followed (New York, the Hague) the A/V working group grew bigger, specifications became clearer and use cases became more concrete. Fast forward to today and IIIF A/V has a long list of “Issues” on its GitHub page promising a great future for the adopters.

But where exactly are we now? We need to understand the difficulty in the challenges that IIIF A/V has to face and why it can’t unroll in the same way the Image API did. Providing access to bitstreams of audio and video content adds a temporal dimension to that of the height and width but more importantly encoding, bandwidth and performance issues come forward triggering a lot of discussion: -One of the first important decisions is the point of processing and choosing where to let the processing weight lie. “Would it be more beneficial to allow most of the transformations to happen in the computing power expense of the client or the server? Should only one be chosen, and which one would that be. If not what should be done and where?”

“Can URL based standardisation be achieved in the same way as in Image API when we want to access A/V content?”

”Should the audio and video channels be supported by the same API?” Questions like that keep coming up, perplexing a much anticipated topic that continues to gain attention. Tom Crane explores the topic strongly in this document.

In January 2017, IIIF issued a Charter for IIIF A/V API clearly defining its scope and a timeline until the end of 2018. At that point the IIIF A/V API is expected to be ready. There’s also a wonderful blog from Jason Ronallo in which he indepthly describes several of the some of the challenges IIIF A/V faces.

Figure 1: Mock up of IIIF A/V presentation API

Many questions regarding the future and feasibility of IIIF A/V remain to be answered. By involving many experts from the audiovisual sector and following the core strengths of IIIF’s development thus far a positive result is expected but costs and resources will remain a high concern for institutions already struggling with digitisation and presentation of their audiovisual content. Ideally, IIIF A/V would depreciate these costs and institute a formula that smaller institutions can follow. This all remains to be seen as IIIF A/V team grapples with the conundrums of the presentation of audiovisual material.

Dr. John B. Howard

Dr. John B Howard is University Librarian at University College Dublin (UCD). He served previously as Associate University Librarian at Arizona State University (ASU) and Professor of Informatics in the ASU School of Computing, Informatics, and Decision Systems Engineering. Prior to that he spent 25 years at Harvard University, where he served as a faculty member in the Department of Music and held a variety Library related positions including: Keeper of the Isham Memorial Library; Richard F. French Music Librarian; Librarian for Information Technology; Director of Digital Library Initiatives; and Associate Director for Informatics and Technology Research at the Harvard Medical School. John has worked extensively in the design and implementation of information and knowledge management systems. These include Digital Repository Services at ASU, the "Digital Antiquity" national archaeological data repository, Harvard's geospatial data library, the Harvard Digital Repository System and associated services, and the Harvard Medical School Digital Library.

Karin Glasemann

Karin Glasemann is the Digital Coordinator at Nationalmuseum Sweden. She was trained as an historian and has worked in digital cultural heritage for almost 10 years. She started in Museum Documentation and has a good knowledge on data- and documentation standards. In 2012, she started working for Nationalmuseum, Sweden, originally to ameliorate digital documentation. However, a major goal of documentation and digitization was soon to make the astonishing collection more accessible to the public. Since 2013 she has been responsible for Nationalmuseum delivery to Europeana, which pushed Nationalmuseums internal development. The Open Glam policy that was effectuated in October 2016 is also due to the collaboration with Europeana. Read more about Karin in her Meet the Members Council Profile!

Themis Karavellas

Themistoklis Karavellas is a scientific software engineer and researcher in the Netherlands Institute for Sound and Vision. He studied Computer Science (BSc) and Information Science (MSc) and is working hard in bridging the domains of Culture and IT.