Digital Preservation Coalition

TIFF or JPEG2000?

We are about to launch on a large digitisation project and at the eleventh hour an issue has come up over which format we should be using. The primary purpose of the project is to provide long term backup copies, although obviously once we have the images we can look at additional functionality.

My question, therefore, is whether TIFF is still the best format for long term preservation (this is our current position), or whether JPEG2000 would be a better option. While the general consensus of responses from my posting yesterday seems to be TIFF, there is an element of horizon-scanning and if JPEG2000 is the next big thing, maybe we should just bite the bullet?

Any thoughts that people have on this today, as my timeframe is quite tight, would be much appreciated!

thanks

Polly Parry

Museum Archivist & Records Manager

Natural History Museum

I would say JPEG2000 has benefits over TIFF (built-in thumbnailing, less variance in specification), but then my preference is to also store the RAW image wherever available. OTOH we're currently stuck with cameras that produce old-fashion jpegs!

Regards

Chris Puttick

Oxford Archaeology

Good morning Polly,

The Wellcome Library is beginning its five year digitisation strategy.

We have chosen to go with JPEG200. We have commissioned a report as foundation to our approach. View that report from our Library website here - http://library.wellcome.ac.uk/doc_wtx052496.html Apologies if you already know about this.  You can follow Wellcome's work with JPEG2000 via our JP2K-UK Working Group website - http://jp2k-uk.wikidot.com/

The DPC have also published a Technology Watch Report on JPEG2000, available from this page - http://www.dpconline.org/advice/technology-watch-reports.html

In our approach we perceive benefits in smaller JPEG2000 file sizes, thus reducing storage costs, with associated abilities to use lossy images whilst still retaining an acceptable level of image quality.

We're also attracted by JPEG2000's ability to use layers and tiling.

We're also clear that this is a digitisation-for-access activity.

Whether JPEG2000 is a suitable format for preservation is a different matter.

Hope this helps,

 Dnt

 Dave Thompson

Wellcome Library

The Tate is reviewing future archive formats at the moment and we have heard mention of JPEG2000 as being a stable and less memory hungry alternative to Tiff. We are consulting at the moment.

David Clarke

Head of Photography

Tate Britain

Colleagues,

I think it is true to summarise one recommendation of the Buckley Report (DPC Technology watch report on JPEG2000 cited by Dave below) as the definition and standardisation of an archival subset of the full JPEG2000 specification prior to its adoption as a preservation format.  This is not a call to use “JPEG2000” per se as a preservation format.  So far as I am aware this has not yet been completed.  Any such subset would need to address the danger of actually destroying information by using lossy compression techniques and no doubt numerous other technological dependency issues in such a complex format specification.  How much of an advantage virtually or actually lossless JPEG2000 offers in terms of extent / verbosity may be a bit of a moot point?

 As far as I am aware, the commitment to JPEG2000 by a few prominent institutions - Wellcome is an example - is limited to use as a surrogate format where the original analogue versions are being retained, i.e. not as a preservation format as such.

It’s probably superfluous to recommend that assessment of any file format needs to take a look at the nine criteria common to most of the literature or at least the five most common ones identified in my recent Technology watch report for the DPC on File formats more generally.......   Archival use of JPEG2000 vs. TIFF needs to take a particularly long, hard look at:

In the TIFF corner:

-transparency (relates to the compression issue above; any compression algorithm needs to be explicit and decodable indefinitely);

-stability (if an archival profile is to be established,  it would need to be managed according to a process like the special procedures agreed by ISO for the stable management and versioning of ISO19005, AKA PDF-A)

In the JPEG 2000 corner:

 -robustness  (having said all the above, I found Volker Heydegger's demonstration of the possible greater robustness of JPEG2000 against bit rot as compared to TIFF at the DPC Report launch event back in 2008 fascinating)

 -interoperability (JPEG is just easier to deliver in whole or part);

 -extent

 plus the more imponderable ………….

 -adoption (your best guess when you’re an early adopter on who is going to go which way when making the same choice).

A repository also needs to consider its overall portfolio of preservation formats in the light of its overall strategy.  If you don’t do that, I’m not so sure the above can really be squared.

Kind regards

Malcolm Todd

Digital Archives Advice Manager

Archives Sector Development

The National Archives