Michael J. Bennett University of Connecticut Storrs, CT/USA & F. Barry Wheeler Library of Congress Washington, DC/USA IS&T Archiving 2010 Conference The Hague, NL June 4 th, 2010

What are they? To a large extent raw file formats are camera vendor-specific and proprietary e.g. Nikon (.nef), Canon (.cr2) Raw file = mosaiced raw sensor data + metadata + jpeg preview Storage of raw image data can vary among uncompressed, losslessly compressed, and lossy compressed options.

What are they? (continued) Peter Krogh, Non-Destructive Imaging: An Evolution of Rendering Technology, (retrieved ), © 2007 Adobe Systems Incorporated. All rights reserved.

an alternative to proprietary raw? Created by Adobe. Currently the only documented and open raw format. Attempts to standardize the universe of multiple proprietary raw formats. An extension of TIFF 6.0. Compatible with the TIFF-EP subset. …more on this later

What are the differences between archiving TIFF vs. raw formats? TIFF files are a fixed rendition (any subsequent post editing = raster image editing = changed pixel data = non-reversible) analogous to archiving a traditional print Raw formats encompass latent image data + metadata which contain processing instructions (any subsequent post editing = parametric image editing = preserved original pixel data = reversible all of the way back to original capture state) analogous to archiving a traditional photo negative

So what? Don’t I have everything that I need in uncompressed TIFF? As time passes, older raw files can be re-processed with newer software without changing pixel data. Such re-processing may be better able to extract image detail, lower image noise, correct for lens aberrations, etc. than the software available at the time of original capture. Raw files, in turn, have the potential to improve over time in a non-destructive manner. Similar post-editing to rendered TIFF files can result in destructive, non-reversable changes in pixel data.

Raw’s advantages vs. TIFF, a summary… 1. access to raw mosaic sensor data 2. capable of benefiting from newer and better demosaicing algorithms over time 3. more control over the original rendering process 4. greater bit-depth 5. wider color gamut options 6. broader usable dynamic range 7. non-destructive parametric editing 8. storage requirements of roughly one third to one half the size of uncompressed TIFF (nearly comparable to lossless JP2)

a summary of advantages… 1. Unlike proprietary raw, DNG maps sensor-specific color space into the standardized CIE XYZ space. 2. DNG is fully XMP-compliant and can use an embedded XMP space for both descriptive and technical metadata. 3. DNG can also write parametric edits to embedded XMP which allows such rendering instructions to not only be completely portable along with the file itself but also allows such instructions to be more easily managed. 4. Accurate embedded JPEG previews based upon parametric editing adjustments are also unique to DNG.

a summary, continued… 5. DNG has the ability to embed custom camera profiles that allow for refined compensation among cameras of the same make and model. 6. DNG furthermore utilizes an internal MD-5 hash for its original mosaic image data. Thus the format is uniquely able to self-validate.

Some Q&A… Q1: How can parametric edits be best protected from alteration when raw files move among recipients and software? A1: (DNG “Snapshots” tag. File last saved date.)

Some Q&A… Q2: What might be lost from the original raw files in converted DNGs? Is the original black point preserved for instance? A2: DNG specification doesn’t require keeping the original black point of proprietary raw files. Adobe’s DNG converter does preserve it, however. In the final analysis converted DNG “fidelity” is dependent upon the chosen converter software.

The bigger picture… Q3: Given the need to reformat over time in an environment of a multitude of capture devices, their unique formats, conversion software packages and their unique OS dependencies, output hardware devices, etc. can DNG allow for better control over this long term ongoing process? A3: The 100 year cultural heritage perspective vs. the current latest and greatest.

The bigger picture… Q4: Given the inability to foresee the future in terms of needed or expected outputs (display, print, etc.) is DNG best poised to meet the broadest possible future requirements? A4: At this point in time, yes.

Michael J. Bennett Digital Projects Librarian University of Connecticut Libraries Storrs, CT/USA F. Barry Wheeler Digital Projects Coordinator Library of Congress Washington, DC/USA