RIT imaging scientist
Roy Berns reveals artworks as they were meant to be seen
|This image shows the
current condition of Vincent Van Gogh's 1890 oil painting,
performed on the faded painting led to this digitized version
of Daubigny's Garden, in which the ground and white flowers
were recolored to simulate the original appearance.
Roy Berns knows a thing or two about Vincent Van Gogh's palette.
He knows time has toyed with Van Gogh's materials and that
the colors we see today are not always as the artist painted them.
Pigments have lost
their intensity, as in the faded ground in his 1890 painting,
Daubigny's Garden. Other colors have all but disappeared.
Berns is giving the
art world high-tech solutions to such conservation challenges.
Berns doesn't restore the works themselves: He has developed
technology that can create color-accurate reproductions for printing,
publishing and for viewing on the Internet. Conservators can use
the technology to detect forgeries and to optimize gallery lighting.
It also can help determine if a painting's surface has been
altered over time and show the affect of the cleaning.
Berns, the Richard
S. Hunter professor at RIT's Munsell Color Science Laboratory
in the Chester F. Carlson Center for Imaging Science, is teaming
up with colleagues in the art world to create the next generation
of imaging technology that will change how museums reproduce and
archive artwork. Berns' system will be the first to document
and reproduce artwork that matches the original under any light
source. He envisions a practical system that will use off-the-shelf
hardware combined with proprietary software, some patent pending.
project, now in its second year, was launched with grants of $315,000
each from the National Gallery of Art in Washington, D.C., and
The Museum of Modern Art in New York City. Berns recently won
an additional four-year grant of $874,000 from the Andrew W. Mellon
Foundation, boosting project funding to $2.3 million, including
$586,000 from RIT.
uses spectrophotometry, a process that measures wavelength and
energy or intensity of light reflected from the surface of a painting.
Data gathered by a spectral camera yields information about the
physics and chemistry of a work of art. Identifying the optical
properties of materials, or an artwork's unique spectral
fingerprint, will bypass some of the obstacles inherent in archiving
art with digital technology.
While the most successful
imaging systems currently used by museums mimic the human visual
system, Bern's research will record the optical properties
of the paint. This is significant because color values change
under different light sources due to the way the eye processes
When we see color,
we see integration, Berns explains. Wavelength information
gets combined into three signals red, green and blue. With
digital photography you get color information, not information
about spectral data, because the camera is working in the same
ways as the human eye.
By using spectral data,
it is possible to calculate an object's color precisely.
will make it possible to measure the color of a work of art rather
than simply estimate it, as is done with film and even the most
sophisticated digital cameras in use today, says Jim Coddington,
chief conservator at The Museum of Modern Art. This increased
color accuracy will open up significant new avenues of research
for museum curators, conservators and art scholars.
Berns believes there
may be commercial applications as well.
I am very excited
about this work, says Berns. I first envisioned such
a project in 1988 and made a pitch to several imaging companies.
Although each was intrigued, the leap to commercial application
was just too great. Today it is clear that visible spectrum imaging
provides many advantages over conventional digital photography,
particularly for artwork. We plan to demonstrate the value-added
of this approach. Who knows? This time around there may be a strong
interest in commercialization.
For more information
about Berns' research, visit www.cis.rit.edu/people/faculty/berns/research.html.