Page Tree:

Child pages
  • Djatoka-based image cropping demonstrator

This wiki space contains archival documentation of Project Bamboo, April 2008 - March 2013.

Skip to end of metadata
Go to start of metadata

Demonstrator Title: Reusing, cropping and referencing digital images
Summary: It's not enough to simply discover relevant digital content -- scholars need to be able to re-purpose and manipulate digital resources (including derivatives and representations of resources), making use of repository-independent tools and being able to reference resources used via persistent URIs and in accord with bibliographic citation standards. This requires that content providers describe and make digital resources available in accord with standards that support interoperability and integration. It requires tool builders to construct tools that can ingest resources without regard to brand or location of repository and can generate output and new derivatives that persist and are themselves referenceable. Additionally, while newer tools and repositories are more sensitive to the need for integration and conformance to emerging standards, additional services and frameworks are needed to glue everything together more seamlessly, which is where Bamboo might come in. This demonstrator focuses on exploitation of digital image resources using a simple Djatoka-based Web application. Djatoka is a robust, repository-independent image manipulation tool through which users and other applications can transform and manipulate Web-addressable digital image resources and create new cropped, rotated, and re-sized derivative image resources with persistent identifiers. The illustration scenario used hints at the potential benefits to be realized when quality digital content and tools like Djatoka are integrated in support of scholarship, but it also highlights the clusimness of current methods used to integrate digital content and tools.

Applicability: This demonstrator is applicable to domains that exploit high-quality digital images which in the course of scholarly use may need to be transformed across image formats, rotated, cropped, and/or presented at multiple resolutions. It anticipates that scholars value an ability to work across disparate discovery portal services and content repositories using repository-independent image and bibliographic tools. It also assumes that such infrastructure components are designed and maintained separately without specific knowledge of each other -- i.e., with a reliance on the Web architecture and complementary standards and best practices. It assumes a requirement for persistent reference to both original sources and generated derivatives. In terms of scholarly practice, it focuses on scholarly activities to do with selecting, manipulating, reusing, and referencing information resources (all supporting the intellectual activities of interpretation and synthesis).

Links: The following resources, tools, standards, and protocols were exploited in creating this demonstrator:

OAI-ORE (a pre-release alpha draft was used)

DLF Asset Action semantics for typing resource representation URIs

American Social History Online (a metadata-based portal created as part of DLF Aquifer)

Zotero and a custom Zotero translator created using Zotero-Scaffold and exploiting UnAPI hooks provided by the DLF Aquifer portal.

Content from the Library of Congress Panoramic Photographs Collection

Djatoka - an open source image-manipulation tool developed by Los Alamos (Ryan Chute and Herbert Van de Sompel)

And in particular an image-cropping application built on top of Djatoka

Among other technologies, Djatoka builds on the OpenURL standard.

Presentation: The simplified, contrived scenario used in this demonstrator postulates a student doing research for a paper which will include a discussion of the tragic 1903 fire which destroyed the Iroquois Theater in Chicago, killed 600 individuals, and resulted in the closing of all theaters in Chicago for a period of time. Through the Digital Library Federation's American Social History Online portal, the student discovers a high-quality digital version of a photo taken the day after the fire. The student wants not only to reference the photo in his research paper (the photo is owned and was digitized by the Library of Congress), he also wants to manipulate the high-resolution, uncompressed TIFF image file to crop and zoom in on a specific region of the photograph to include in his paper. He wants to be able to provide URIs for both the cropped region of the photograph (i.e., the image fragment) he creates and for the source Library of Congress image from which the image fragment he puts in his paper was derived. He does not have on his desktop any software that can manipulate the original TIFF, nor does he have a Web server at hand on which he could mount the generated image fragment, even if he could create it on his own. By exploiting the technologies mentioned above and by some mildly tedious manual cutting and pasting, the student is able to achieve the desired end-result; however, clearly the process could be more seamless and more automated. A five and a half minute video clip has been created showing how the student might go about his task using currently available technology and content. Formats:

For Windows Media Player (.wmv, 8 MBs, embedded player)

For QuickTime Player (.mov, 12 MBs, embedded player)

For Real Media Player (rtsp:// (.rv, 9 MBs, streaming)

Discussion: Notice from the video that the student does not have to handle the TIFF file directly -- interaction is between the Library of Congress repository and the Djatoka-based image manipulation application bypasses the student's desktop once initiated. Notice in fact that the student does not even have to visit the Library of Congress Website himself -- in this instance, sufficient information to support the required interaction and referencing tasks is already present in the DLF American Social History Online portal. The Djatoka-based application not only allows the student to manipulate the TIFF image and create desired derivative using simply a standard Web browser, it can also be configured (through caching) to serve as a de facto repository for this new digital artifact (depending on local service implementation policy).

However, while the made-up scenario outlined above can be demonstrated today for a selection of digital content on the Web, a lot of things have to work just right. Many digital content providers do not provide asset action URIs (or the equivalent) pointing to best resolution image files they hold, or do not disseminate same in an OAI-ORE Resource Map. Nor is most digital content described by rich, MODS bibliographic records (facilitating reference in accord with bibliographic standards). The Zotero translator used in this illustration is custom, experimental and not in wide circulation. And while the Djatoka tool as available for download from Los Alamos is quite robust and can do much more than just crop and zoom, it may be useful to create additional service wrappers to help integrate this tool into additional scholarly workflows. In particular, there seems good potential for benefit by implementing a service layer for batch processing of aggregations or collections of images using Djatoka. Additional standards and best practices are needed to encourage more scholarly high-level exploitation, manipulation, and re-use of digital images.

This is where Bamboo could come in. The scenario illustrated here suggests that:

  • In addition to providing a URL to a human-readable splash page for each resource held, Bamboo content partners should provide persistent URLs for representation(s) of each resource (and potentially resource components) amenable to analysis, manipulation, re-use, re-purposing by scholarly tools and services. Such URLs should either be disseminated in accord with community standards and best practices (e.g., ATOM, OAI-ORE, ...) or should be made visible from splash pages such that automated resource URL brokering services can "discover" these additional URLs through spidering and associate them as added metadata with base resource. Bamboo could build such a URL brokering service for actionable URLs that would facilitate access by scholarly tools to, in this instance, high-res representation of digital image resources.
  • Personal collection / personal library management tools (e.g., Zotero) should be able to ingest and disseminate not only descriptive metadata about resources of interest, but also metadata about associated components and representations of a resource that a scholar might want to submit to a tool in order to analyze, manipulate, etc. For Zotero, the ingest part might be accomplish through implementation of translators (as in this illustration). These could be developed and promulgated by Bamboo.
  • Tools designed to facilitate scholarly interaction with digital resources should be able to ingest digital resources via URL.
  • Tools should provide human and machine interfaces consistent with Web architecture priniciples -- e.g., so users can interact with tools via their Web browsers.
  • Tools should generate Web-referenceable and preservable representation(s) of output -- i.e., so that tool output can be ingested into repositories and so become digital resources in their own right.
  • No labels

1 Comment

  1. Unknown User (


    This looks like a good scenario to work with -- well-framed problem, and a good use of real tools. I particularly liked the fact that you took the photo fragment all the way into the word document and did the citation -- I think completing the story in that way will improve people's ability to connect with the technology. My primary suggestion for improvement is to work some of what you supply in the description into the video presentation -- if I were a naive scholar just watching your video, I might miss the areas where Bamboo can help, and therefore think that all of the work had already been done. Emphasizing the "if an image service existed, I could...." portions would handle that, I think. My other thought is perhaps useless -- but, do you think it would be better to phrase the demonstrator in terms of a faculty member preparing a journal article? Or would that change the scenario too much? I imagine that people can extrapolate between the two, but since the standards for research publication are higher, it might be useful to show the demonstrator in those terms.

     It looks like you have this very well in hand. Can I do anything to help?