Cultural Heritage photography is often dealing with vulnerable materials. Some - like film - may deteriorate over time. Digitization is crucial to preserve it for the future and guarantee access without exposing the originals. Ideally the quality of the digital copy is good enough to replace the original. The FADGI and METAMORFOZE initiatives are attempts to give clear guidelines that can be objectively verified.
While striving for excellence is honorable, not all materials require the same amount of precision. Digitizing a famous artist's historical handwriting requires a higher quality level than the archives of a newspaper. In both cases text recognition is equally important. The handwriting's file however needs to show the structure of the paper and have a tonal range resembling the original as close as possible. Qualities you may forego when digitizing a newspaper. Therefore both FADGI and METAMORFOZE offer various quality levels. FADGI uses stars to distinguish its levels. FADGI**** is the highest obtainable quality with today's technology. METAMORFOZE has three quality levels, which roughly correspondent with FADGI two, three and four stars. The FADGI one-star level has no equivalent in METAMORFOZE. It's for localization only, like a thumbnail in a library.
In some publications ISO Standard 19264-1 is mentioned. It's distilled from the FADGI and METAMORFOZE results.
METAMORFOZE was initiated by the Dutch National Library and the Dutch National Archives. Therefore it focuses solely on the digitization of printed materials. Guidelines for translucent materials are missing. The Federal Agencies Digitization Guidelines Initiative (FADGI) had a broader scope. For that reason we'll look at FADGI below.
Requirements
FADGI provides guidelines for the digitization of a multitude of materials, ranging from rare bound volumes to photographic transparencies to radiographic X-Ray film. For all these materials it quantifies:
- File Format Resolution
- Bit Depth
- Color Space
- Tone Response
- White Balance Accuracy
- Illumination Uniformity Color Accuracy
- Color Channel Misregistration
- MTF Requirements
- Scale Accuracy
- Maximum Sharpening
- Maximum Noise
Resolution is an easy one; it can be calculated without requiring any specific measuring tools nor software. For a FADGI**** digitization of photographic negatives from 35mm through 4x5" a resolution of 4000 ppi is required. Photographic prints require 600 ppi to adhere to the same standard.
The sensor of a GFX-100 offers 11,648 x 8,736 pixels. This theoretically enables us to digitize a 24x36mm negative at a resolution of 11,648/36 x 25,4* = 8218 ppi.
*25,4mm fit into one inch
An 8x10" print: 8,736/8 = 1092 ppi. Since the proportions of the original materials do not always match the sensor's we're calculating with the least favorable number. You don't want to crop your material.
For the originals mentioned above, the GFX easily meets and exceeds the highest FADGI standards. At least when it comes to resolution. When we're looking at a larger film format like 21/4 x 21/4 the result of 3882 ppi falls short to meet the highest standard. This would be a FADGI***.
The numbers only tell us what resolution a GFX-100's sensor theoretically is capable of. They don't mean anything, when the light hitting those pixels had to pass through a mediocre lens. A simple calculation won't do. Other tools are needed to provide objective standards.
As providers of digitization hardware we can give our partners some of the tools needed to meet the desired targets. Hardware and procedures provided by other vendors will always be needed to create a complete solution. Sticking a batch like "FADGI Compliant" to a product is therefore pretty meaningless, unless you're selling a closed system, like a scanner or a boxed solutions with an integrated camera. Due to the way they're built these systems generally are suitable for a limited set of materials and sizes. A universal system handling all types of materials and image sizes in a closed plug-and-play environment doesn't seem feasible.
Measurement is crucial
Certification for FADGI compliance requires objective testing methods. These are outlined in the Digital Imaging Conformance Evaluation program (DICE). A DICE procedure always requires:
Reflective and transmissive image targets
Software to analyze the captured targets
A Device Level Target (DLT) takes the place of the object that's to be digitized and fills the best part of the frame. It includes color and grey patches, a ruler and patches to evaluate sharpness and to detect aberrations. A DLT is typically captured at the beginning and end of a batch.
An Object Level Target (OLT) basically features the same patches as a DLT but is small enough to be captured alongside every object. Whereas a DLT is easier to evaluate thanks to its larger size, some institutions oblige their photographers to include an OLT in every capture.
Software such as provided by Image Engineering and Image Science Associates enables objective analysis of DLT and OLT captures. The only way to know whether a certain FADGI or METAMORFOZE claim is valid or not.
Universal Test Target (UTT) is a generic term which can refer to both DLT and OLT.
The FADGI Technical Guidelines (page 10-12) give detailed information about the various parameters to be measured in a DICE procedure. This report reviews and explains all points mentioned under requirements and can be downloaded free of charge.
How we can help
As outlined above, FADGI helps us to quantify most of the parameters of the digitization process. It also pays attention to Best Practices. Although these might be less measurable, they are equally important. In general there's a lot of knowledge exchange between people working in this industry, which makes Best Practice more than just a hollow phrase. It's very well possible to create FADGI compliant images without adhering to technical recommendations and best practices. However, it may become a lot easier to maintain an efficient and repeatable workflow when you do.
Here Cambo can play a role. Products like our RPS copy stands and film stage provide the necessary stable platforms. A stand like the MST can mean the same in the digitization process of works of art that are too large for flat lays. It would be useful to create a comprehensive set of lens/camera combinations that give potential users a clear insight in what's achievable - and what's not. For film digitization e.g. we could create a table showing which standard a certain camera/lens combination can achieve for various film sizes. Ideally we would limit ourselves to two or three lenses and have the results evaluated in the appropriate software. A Fujifilm GFX- 100 attached to a Cambo repro camera with a Rodenstock Digaron 105 Float would certainly be capable to adhere to FADGI**** for the digitization of 35mm film and FADGI*** for 120 roll film. The latter would require the larger Phase One sensor to meet FADGI****. Anyhow, if we phrase it that way our FADGI claims would be verifiable and help potential clients to make a decision and justify funding.