DZI to IIIF?

[Robert Sanderson]

Has anyone written a shim that converts IIIF URLs into DZI requests?

Or any takers to do it? :)

Rob

--

Rob Sanderson

Information Standards Advocate

Digital Library Systems and Services

Stanford, CA 94305

[Jon Stroop]

In fact...

https://gist.github.com/jpstroop/4624253

this is still to IIIF 1.0, but not a big deal to update.

This was gutted from before OSd spoke IIIF.

--
-- You received this message because you are subscribed to the IIIF-Discuss Google group. To post to this group, send email to iiif-d...@googlegroups.com. To unsubscribe from this group, send email to iiif-discuss...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/iiif-discuss?hl=en
---
You received this message because you are subscribed to the Google Groups "IIIF Discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to iiif-discuss...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Tom Crane]

(Batch conversion of .dzi pyramids to level 0 IIIF pyramids)

In a similar vein, and starting with this code, has anyone written a utility that takes as input a dz image pyramid (.dzi xml + tile directories) on disk, and generates an info.json file and a renamed set of tiles (i.e., saving new IIIF static tiles to disk)? This would allow people with static dz resources to see them in IIIF viewers, and reference them in manifests, without requiring new server infrastructure (i.e., anything more than a web server). Would that be a level 0 profile? 

Optionally the tool could take the path of a thumbnail that it could add to the info.json as the "thumbnail" property. The tool would need a base URI parameter too.

I suspect that this tool does not exist - but it would it be useful to collections that already have static assets on disk in this format, and give them a taster of IIIF, without requiring any new server side infrastructure.

It would also allow "home users" to use a tool like Deep Zoom Composer (which outputs a dz pyramid). 

Happy to have a go at creating it - but better check first!

Tom

[Jon Stroop]

Nothing that I know of, but just to suggest an approach--what if (perhaps optionally) this tool-to-be-built just made symlinks that support the IIIF canonical syntax? It might some folks ease into migration.

[Tom Crane]

Started taking a quick look at this and ran into an immediate problem. All the seadragon examples I have that had been generated from Microsoft's Deep Zoom Composer tool have .dzi xml like this:

<?xml version="1.0" encoding="utf-8"?>

<Image TileSize="256" Overlap="1" Format="jpg" ServerFormat="Default" xmlns="http://schemas.microsoft.com/deepzoom/2009">

<Size Width="4608" Height="6136" />

</Image>

The images on disk are all 258x258 tiles, they overlap by one pixel.

Is there a way of dealing with that in the info.json alone?

Deep Zoom Composer is a nice, free tool for ad-hoc image pyramid generation (Windows only though, I think). You can lay out source images, and it can export a whole static web site ready to run, using Seadragon Ajax (ancestor of OpenSeadragon). It would be great if one extra batch process could turn that into a IIIF image pyramid + manifest, ready to load into [insert IIIF viewer here]. 

A process that has to shave pixels off the tiles isn't really what I was after - though it's still a possibility.

Tom

[Robert Sanderson]

We don't currently have a way to specify overlap, as per the DZI attribute, so we can't describe that tile set in an info.json document.

Two questions:

* Should we?

* And if yes, shall we add it as a ticket to discuss next week, or on tomorrow's call?

Thanks :)

Rob

--
-- You received this message because you are subscribed to the IIIF-Discuss Google group. To post to this group, send email to iiif-d...@googlegroups.com. To unsubscribe from this group, send email to iiif-discuss...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/iiif-discuss?hl=en
---
You received this message because you are subscribed to the Google Groups "IIIF Discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to iiif-discuss...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jon Stroop]

* Should we?

Many years ago I used this application[1] to generate a pile o' tiles using the DZI syntax. It parameterizes the tile overlap[2], and after some experimentation I wound up setting it to 0. I could certainly be missing something, like a more complex use case,  but I never really saw the point of the overlap and it didn't seem to make any difference to Seadragon at the time.

Does deep zoom composer let you change the overlap, by any chance?

-Js


1. https://github.com/openzoom/deepzoom.py
2. https://github.com/openzoom/deepzoom.py/blob/develop/deepzoom.py#L80-L81

[Tom Crane]

>> Does deep zoom composer let you change the overlap, by any chance?

Not that I can see, unfortunately.

>> * Should we?

There are two reasons why this might be good to have:

1) Collections that were "sliced" a while ago. http://cudl.lib.cam.ac.uk/collections/newton has dzi with an overlap of 1, although I think they sliced theirs with a Perl script. While externally hosted shims can convert them to IIIF, they might not be averse to IIIF versions sitting side by side on their own servers, if they are just static files.

2) Deep Zoom Composer is a friendly end-user tool. It makes it easy for home grown deep zoom projects - if one extra conversion step lets people pick their favourite from the soon-to-arrive plethora of IIIF viewers, it helps drive adoption, right down to the "school project" level.

Neither or these two reasons may be worth the effort though. It's not a pressing requirement for me, it came about from an enquiry. The overlap problem feels like it would require an awkward addition to the model - though obviously it should be easy for OpenSeadragon to deal with it as it already does it for dzi.

* And if yes, shall we add it as a ticket to discuss next week, or on tomorrow's call?

Maybe a quick mention to see if it's worth pursuing. How many static dzi collections are there out there?

Tom