Direct Digital Image Capture:
Report of a Workshop
As institutions switch from film to digital photography,
how can they produce the most accurate, reliable and useful images of
the works in their collections? This was the central question at a recent
conference at the Rochester Institute of Technology. Over one hundred
representatives of nearly fifty institution and nine companies met September
21-24, 2004 at R.I.T. to hear presentations, see demonstrations, and
discuss this question. This article is a report on and critique of that
To document current practices, a survey was conducted of digital imaging procedures currently used at American Museums, Libraries, Archives, and other cultural institutions. Over forty institutions submitted answers to the questionnaire. Because no comparable survey has been conducted, the results should be of considerable interest to cultural institutions, indeed to anyone interested in high quality digital photography.3
The survey forms were to be filled out by those responsible for the photography department at [each] institution. The survey forms included 77 questions in all, including detailed questions regarding the personnel involved, training and attitudes regarding digital photography, imaging workflow, and about each institution's specific practices in the use of digital photography, such as studio setup and equipment, image editing, color management, and maintenance of image files.4 As one might expect at this early stage in the adoption of direct digital photography, there was considerable spread in most of the answers. Some of this spread undoubtedly resulted from institutions having used more than one system or software package as new options became available. This diversity is surely not all to the bad, since the more real-life experiences on which we have to draw the better. Nevertheless, it was surprising to find so many color management software packages having been used and such a range of hardware and software used for calibration of the display. It was a bit disturbing to see that different institutions reported such different points at which initial sharpening first occurred in their workflow (12 institutions before printing, 9 other, 6 reporting highest quality digital master image is sharpened before saving, 3 at capture, and 2 do not know). Likewise, it was disturbing to see that only 49% of responding institutions archive the original raw file from the camera; that, of the responding institutions, 1 stored the digital masters in GIF file format, 5 in Photoshop, 7 in JPEG, and only 39 in TIFF; and that 8 of responding institutions backup their images on tape, 8 on DVD, 11 on CDs, and only 12 on other systems (presumably including the most archival storage, servers and hard-drives).
Perhaps of more long-term importance than these recent technical differences were the different approaches noted by responding institutions. Only a few of the questions focused on differences in approach, but these answers were surprisingly varied. In answer to the question, Which of the following statements best describes your definition of image quality?: 46% selected the image on the screen looks like the original; 15% selected specified values for a target are matched in the digital file; 8% selected the measurements on the output device match the measurements on the original; 8% selected you have an aesthetically pleasing reproduction on the display; 8% selected you have an aesthetically pleasing printed reproduction; and 15% selected other. In the discussion also, it was clear that some institutions archive digital images primarily for publication in books or other formats, while others believe that the archived image should be use neutral.
For some of
us, the choice most in need of further investigation is the complex
question of how the various types of reproductive images can most accurately
look like the original. In the questionnaire, there were
three questions about the type and placement of light for viewing reproductive
images, but none about the conditions for viewing the original works
of art. In recent years, lighting of paintings in museum galleries and
photography studios has been quite flat.
In addition to the survey, site visits were conducted
by the R.I.T. team to six major early digital image adopter institutions,
with return visits to four.5 Although answers
to the questions regarding studio set-up and equipment were not included
in the printed results distributed to conference participants, these
were described in presentations by the R.I.T. team and by representatives
of the site-visit institutions. Here again the range of studio set-ups,
lighting, cameras, and lenses was surprising. Our first reaction might
be to question the lack of consistency among institutions, but, at such
an early stage in direct image photography by museums, the range of
practice has been natural and the range of experience informative. One
large, well funded museum has purposely purchased a variety of cameras
and lenses as major changes have taken place and has tried out different
lighting and copy set-ups as a way of becoming familiar with the options
and making informed decisions.
noted at the visited institutions also depended on the history of digital
adoption at each institution and on the uses for which digital photography
has been carried out. In some cases, institutions began digital photography
for small, individual projects; in others as a large campaign to digitize
entire collections. In some cases, digital photography is carried out
specifically for posting on the web, in others primarily for print publication
in catalogues, in others primarily for high quality archiving, in others
largely for collection management.6 At one
institution, at least, the specific use for each shot is identified
and the digital photography is carried out to achieve this. Of course
decisions have also been influenced by the nature of the material being
photographed. Libraries are much more likely than museums to batch process
images after studying a representative sample.
The R.I.T. team
performed various systematic comparisons of photographic methods at
the visited institutions, attempting to quantify each comparison. These
were described by Erin Murphy and will be included in her master's thesis,
currently being completed at the Munsell Color Laboratory at R.I.T.7
Perhaps the key comparison presented and available for inspection in
the laboratory was a comparison of the digital images produced by each
of the four revisited museums from two monitor size, multicolored paintings,
each institution using its standard equipment and procedure. The colors
of the standardized prints of the four resulting digital images varied
enough in color to make the R.I.T. team and other conference participants
concerned about the color accuracy of current museum procedures. Less
surprising, for both paintings the color of all four of the prints varied
somewhat from that of the original paintings when viewed next to each
other. One should note that the visited institutions were full participants
in this process, each welcoming the R.I.T. team members for a full day
of tests and experiments.
A central goal of the R.I.T. project is to devise a way to produce more consistent and more accurate images of works of art, which would also allow institutions to dispense with the costly and time-consuming process of post-capture visual editing. To pursue this, Roy Berns and his colleagues have been experimenting with spectral imaging, identifying the optical properties of a painting's materials, representing a work of art by its spectral data rather than how it looks to the viewer or to the camera. Because of the limitations of current museum practices, a major project to develop a spectral imaging camera is underway, co-sponsored by R.I.T., the National Gallery of Art, Museum of Modern Art, and the Andrew W. Mellon Foundation.8
Original Works of Art
It is important to note that this highly promising spectral-art-imaging
project aims to make possible more accurate images of original works
of art, not simply more accurate images of printed reproductions or
other images. This is, of course, a vastly more challenging undertaking.
The aura of
the original work of art has become an almost mystical concept.
In the final analysis, there is surely an aspect of this experience
that cannot be explained; but some essential characteristics of original
works of art put special demands on any attempt to achieve their direct
digital capture. Unlike reproductions in books or on monitor screens,
original works of art have highly complex surfaces, and these complexities
convey important aspects of each work's meaning. Even if we set aside
architecture, with its size and encompassing volume, and sculpture,
with its complex 3-dimensionality, we need to recognize the ways in
which the 3-dimensionality of paintings makes for an experience different
in kind from that of reproductions. The term direct digital capture
is perhaps misleading, suggesting as it may a digital replica.
As we view paintings
(the focus of the R.I.T. project), we do not hold our heads still or
move them from left to right, in and out, but in an unconscious fluid
motion, exploring whatever we find of interest: the astonishing detail
in a renaissance portrait, the gently moving clouds in an impressionist
landscape, the vigorous energy of a Picasso abstraction. The inquisitive
nature of this viewing process, for pleasure and discovery, and in the
case of research often to answer specific questions, is central to our
experience of art. If the painting is small, we may even hold it in
our hands, turning it near a window, to examine it more thoroughly.
As we view a painting in this way, our angle of view changes, as does
the angle of light falling upon it, altering the colors and reflectivity
of the different paints and revealing the expressive impasto of the
painting's surface. The artist's hand comes alive, allowing us to relive
the process of creation and to feel the artist's response to his subject.
In viewing reproductions
of paintings, even reproductions of small paintings that can be reproduced
exact size, even reproductions of the highest quality with near facsimile
color, the complexity and expressive quality of the surface is largely
missing. Several institutions have been carrying out research to find
digital equivalents for these characteristics of original works of art,
but the challenge is much greater than the ability to record the optical
properties of a painting's material, remarkable as that achievement
One aim of the
R.I.T. project is to develop procedures to quantify image quality.
But the concept of image quality depends on the information
we desire to record, the purpose of the imaging, and the uses to which
the images will be put. At the conference there was an especially instructive
discussion about the background desirable for photographers at art museums.
Some favored hiring technical specialists, though others felt that a
background in commercial photography and lack of an art background resulted
in sterile images, insensitive to the individual works of art. Some
favored hiring those with art backgrounds, though others felt that this
turned the images into new works of art.
At the conference,
representatives of a number of institutions described the importance
of capturing and archiving extremely high quality images of each painting
only once, to provide for all future derivative uses, so that the costly
and time-consuming digital capture process would not have to be repeated.
Although this has been an effective argument in obtaining funding for
digital projects, it seems to many of us an impossible goal. No matter
the quality of any current digital images, experience teaches that there
will be new types of digital imaging, different not only in quality
but perhaps also in kind, that will become the new standard for high
quality reproduction and advanced research.
In spite of
abundant evidence to the contrary, the concept of a single iconic image
of each painting pervades the publishing world and to a large extent
even the teaching of art history. But not only do works of art look
different when displayed in different environments with different lighting
(underscored by the R.I.T. team and accommodated for in their tests)
but works of art themselves change. Even if we again set aside architecture
and sculpture, exposed to the elements and other intrusions, paintings
also change, more continuously and more significantly than is recognized
outside the conservation world. Paintings are damaged and deteriorate.
Paints fade, colors change, different colors change in different ways
altering the relationships among areas of a painting, dirt accumulates,
varnishes darken. It remains common practice for paintings to be cleaned
and touched up as they pass through commercial galleries, in accordance
with the taste of the time, to make them more saleable. Paintings are
restretched, relined, and reframed. Museums clean and restore their
paintings. For museum records and research, it is necessary to record
these changes, and for exhibition catalogues it will be necessary to
publish images comparable to what is displayed on the walls. Re-photography
will continue to be a necessary fact of life.
The sharing of current practices in the R.I.T. survey and on-site tests and the sharing of ideas at the conference provide, for the first time, reliable information about the state of high quality digital photography at American institutions. It is clear that commonly held benchmarks do not currently exist. It is also clear that institutions have different concepts of what they hope to achieve when imaging their collections. Institutions are attempting to address these differences, and a major project to develop a spectral imaging camera is underway. Much work remains to be done to achieve images that convey the types of information available when viewing original works of art
1. The conference was titled American Museums Digital
Imaging Survey Benchmarking Conference: Direct Digital Image Capture
of Cultural Heritage in American Institution. A brief description
of the conference, with links, is available at http://www.cis.rit.edu/museumSurvey/conference.html.
2. See http://www.cis.rit.edu/museumSurvey/research.html.
For a report on a closely related conference, also sponsored by the
Andrew W. Mellon Foundation, and the National Science Foundation, see
3. Although the survey period is now over, the survey
questions can still be viewed at http://www.cis.rit.edu/museumSurvey/survey_noSubmit/intro.php.
The survey is titled Direct Digital Capture of Cultural Heritage:
Benchmarking American Museum Practices and Defining Future Needs.
The finished report, including compilation of answers to the survey
questions, is expected on-line and in print toward the end of 2004.
4. Although this review does not include consideration
of administration matters such as staffing and funding, it should be
noted that, at the conference, there was widespread agreement that the
single greatest imaging need at most institutions was an institution-wide
image management system.
5. Site visits were conducted at the Art Institute of
Chicago; Harvard University Art Museums, Harvard University Libraries;
Metropolitan Museum of Art; Museum of Fine Arts, Boston; and the Museum
of Modern Art.
6. The R.I.T. project and conference focused on digital
imagery and were not intended to address the comparison with film-based
photography. For example, in the survey questionnaire, question 19 read
Has your institution started using digital photography? If NO,
you have completed the survey; thank you; without asking why
not? Nor were there any questions about how institutions were
handling their film based photography, although the director of photography
at one major museum stated that the most important recent step taken
by their museum for photography was to put their film based photographs
into cold storage. A few advantages of film based photography were mentioned
at the conference: the advantage of retaining a hard copy back-up (especially
desired by conservators), the convenience of having all material on
each object in a single file folder (especially desired by curators),
and the much less costly process of archiving film based photographs
with link to a full-text PDF, documenting comparative storage costs).
There was some discussion of the financial advantages of converting
entirely to digital imagery versus maintaining the option of film-based
photography where it retained advantages for some museum functions.
7. Erin Murphy's thesis is titled "A testing procedure
to characterize color and spatial quality of digital cameras used to
image cultural heritage." Her March 15, 2004 document A Review
of Standards Defining Testing Procedures for Characterizing the Color
and Spatial Quality of Digital Cameras Used to Image Cultural Heritage is available as a full-text PDF.
8. See http://www.cis.rit.edu/museumSurvey/news.html,
with link to a pdf announcement.
9. For architecture, see for example http://www.cvrlab.org/research/research.html#publications.