Infrared digital cameras

Ian McClure <ipm1000 [at] cam__ac__uk> writes

>We are considering the purchase of an infrared enabled digital SLR
>camera, to photograph paintings when infrared reflectography mosaics
>cannot be justified.

Regarding the astronomical version of the Canon EOS 20D, (the EOS
20Da), Canon emphasizes that their filtration modification is
designed to enhance sensitivity to the 656nm hydrogen alpha emission
band (Ha), which is strongly emitted by gasses in nebulae. It is
likely that this filtration is band-pass around 656nm rather than
long-pass with transmission at 656 and longer wavelengths. This
design is typical for Ha filters used for astrophotography, as it
eliminates visible light pollution as well as the strong natural IR
emissions of the night sky between 700 and 860nm, both of which will
diminish the contrast of images of nebulae. Confirming the
likelihood of band-pass filtration is the relatively minor effect on
natural color rendition in non-astronomical images taken with the
camera; materials highly reflective in the near IR just show a
slight magenta or reddish cast. The 20Da thus appears not very
suitable for digital infrared photography in conservation. It should
be noted that spectral curves of the IR blocking filtration in
standard Canon digital SLR's show no transmission above 700nm,
making these cameras unsuitable for any IR photography.  This
differs from the filtration in standard Nikon digital SLR's, which
generally transmits a sufficient amount of IR above 700nm (and
likely, at least out to 900nm) to permit relatively good IR
photography.

To stay within the Canon family, a custom modified camera, however,
should provide excellent results.  In addition to those provided by
<URL:http://maxmax.com/>, I would look into those provided by Hutech
Astronomical Products
<URL:http://www.sciencecenter.net/hutech/canon/> preferred
by many astrophotographers. They provide modified versions of
several Canon SLR's, including the 20D and the new full-frame 12.8MP
5D, and the 12.3MP Fuji S3Pro. They also will modify customer-owned
cameras in these models as well as Nikon D70's and D50's, although
no warranty is provided (their cameras have a 1 year warranty).
They have a selection of filtration options. For conservation
purposes, order the "clear" rear filter (over the CMOS) which
permits transmission from 380nm to beyond 1000nm, or the limit of
sensitivity of the array.  This filter retains the ability to use
the camera's autofocus system; which would be lost if the original
IR blocking filter was just simply removed.  To use the camera for
normal visible light photography, the Hutech "VLC" filter would also
be needed.  It serves as a temporary replacement for original IR
blocking filter. Hutech provides it in two types of mounts, either
behind the lens or a standard front lens mount.

Just as with film, the question of resolution and format size should
also be considered.  The small format 6 or 8MP cameras will provide
resolution equivalent to, or only slightly better than 35mm IR film;
thus images of larger paintings with fine underdrawings would very
likely still need to be mosaicked.  A practical single-capture,
minimum resolution standard for optimal viewing of fine lines in
underdrawings is that sufficient to produce a 1:1 or actual-size
output (the same basic standard is equally appropriate when
calculating file size for mosaicked images). For photographic
quality 300dpi hardcopy output, a 6MP (3000 x 2000 pixel) camera is
sufficient to record a 7 x 10 inch area of painting surface at 1:1,
and thus for publication or for hardcopy records, is quite
satisfactory for recording selected significant passages in most
underdrawings.

Creating a hardcopy printout of the entire surface of a painting at
full-size is, of course, impractical for paintings of any
appreciable size; but viewing the entire surface at full size (and
larger) is quite practical when viewing on a monitor.  The minimum
single-capture resolution needed to produce a conservation record
that will allow for 1:1 viewing on a monitor is easy to calculate.
For a typical 12 x 9 inch (15 inch diagonal) monitor at XGA
resolution (1024 x 768), the linear resolution is 85dpi (1024/12 or
768/9). Thus by multiplying the dimensions (in inches) of the
painting by 85, you can calculate the number of pixels needed in
each direction. By multiplying these together you get the total
number of pixels needed in the array as well as the file size, in
bytes, of your IR photograph (assuming a single-channel 8 bit file).
A 51 x 34 inch painting, for example, would thus require a 12MP
(4350 x 2900) array for a single capture image without mosaicking;
and a 6MP array would be sufficient for 35 x 24 inch painting (35 x
85 = 3000; 24 x 85 = 2000, approximately).

For best results with large surface area artifacts like paintings,
however, it probably would be best to think again in large format
terms.  You might wish to consider as a replacement for your old
Phase One, a Better Light scanning back for this purpose.  They can
be purchased for as little as $9500 to provide an equivalent 48 MP
resolution.  By the rule of thumb above, this resolution is
sufficient to record the entire surface of an 8 x 6 foot painting
for viewing at actual size on a monitor, or to produce a 27 x  20
inch, 300dpi hardcopy print, without mosaicking. We have had a
Better Light 6000-2 for the past 5 years.  Our students find it very
easy to use and it has been extremely reliable. We use the back not
only for our standard documentation, for which it is outstanding,
but for reflected and transmitted infrared photography, photography
of infrared luminescence, and false color IR imaging.  It has
excellent IR sensitivity out to 900nm, which is fully advantaged
because the IR blocking filter required for color photography is
external and removable; thus scan times rarely exceed eight or ten
minutes with apertures as small as f22 (87C filtration; 170fc
tungsten illumination; 1200 ISO, images are virtually noiseless).
The Better Light back also provides an electronic focusing aide,
which, of course, is particularly helpful in infrared photography.
The manufacturer is strongly centered on applications involving the
photography of works of art, and <URL:http://www.betterlight.com/>
provides much useful information on topics from metamerism to the
safety of photographic light sources in terms of artifact exposure.

In terms of IR sensitivity, the CMOS or CCD arrays in digital
photographic cameras generally have no sensitivity beyond 1000nm
(The IR sensitivity of Kodak IR film, which exceeds that of all
other IR films, reaches 900nm). Sensitivity out to 900 or 1000nm,
however, is sufficient to render most underdrawings clearly, but is
not optimal for penetration of some pigments such indigo, azurite,
raw sienna, and lead-tin yellow, which are more effectively
penetrated above 1500nm.  In those rare situations where these
pigments need maximum penetration, IR imagers must be used.
Unfortunately, the maximum resolution commonly available in these
cameras is only 0.3MP (640 x 512 pixels) sufficient for 1:1 monitor
viewing of an area 7.5 x 6 inches or a 1:1 hardcopy printout of an
area only 2 x 1.7 inches, thus making mosaicking essential if they
are required.

A general note on terminology: If imaging between 700 and 1000 nm
(the type of imaging discussed above), proper terms are "reflected
infrared or transmitted infrared digital photography or photographs"
(omit "digital" if using IR film). If imaging above 1000nm and up to
3000nm, the proper terms are "infrared reflectography" or "infrared
transmittography" and "infrared reflectogram" or "infrared
transmittogram". This is in keeping with the definitions set down by
J.R.J. van Asperen de Boer who first published the terms infrared
reflectography and infrared reflectogram . Above 3000nm, terms are
"thermography" and "thermograms."

Dan Kushel
Distinguished Teaching Professor
Art Conservation Department
Buffalo State College
1300 Elmwood Avenue
Buffalo, NY 14222


                                  ***
                  Conservation DistList Instance 19:29
                 Distributed: Tuesday, December 6, 2005
                       Message Id: cdl-19-29-008
                                  ***