ORIGINAL PAPER

An inexpensive system for continuous lake core photography

Kyle McMillan

Received: 14 December 2007 / Accepted: 8 May 2008 / Published online: 30 May 2008

� Springer Science+Business Media B.V. 2008

Abstract Representative images of split sediment

cores document geological records, serve as visual

aids for multimedia, and provide samples for image

analysis. Existing core imaging systems, often con-

taining integrated hardware and software, are capable

of providing excellent images; unfortunately these

systems are also bulky and expensive, and are

therefore limited to labs that process large volumes

of core and those actively involved in image analysis.

For most research groups such systems are unfeasible

to buy and maintain. Producing good quality core

images with a hand-held digital camera is very

difficult without a consistent lighting source, and

changes in camera angle (in all three spatial dimen-

sions) between photographs may prevent accurate

image compositing. Tripods allow for camera stabil-

ity, but typically do not accommodate downward-

facing photography. Presented here is an easily

constructed core-imaging system that minimizes

many of the drawbacks of personal digital cameras.

Software necessary for using this system is readily

available and can be run from a personal computer.

Keywords Imaging � Core � Photography

Introduction

Core images are increasingly being used, qualita-

tively and quantitatively, as records of lacustrine

paleoenvironments (e.g. Cooper 1997; Petterson et al.

1999; Saarinen and Petterson 2001; Francus et al.

2005). However, many research groups that work

with lake cores do not adequately image their cores

because they lack the necessary equipment and

expertise to do so. The system described in this

paper can be easily constructed and operated, and is

designed to produce high quality core images that

will meet the needs of many users. However, this

system lacks some features that may be found in

more advanced core-imaging systems. For details on

more advanced imaging systems, readers should

consult technical publications and resources from

labs that are equipped with such systems, such as the

Limnological Research Center at the University of

Minnesota, Minneapolis (http://lrc.geo.umn.edu/),

and labs aboard Ocean Drilling Program ships (e.g.

Merrill and Beck 1995; Nederbragt et al. 2000).

The system presented in this paper consists of both

hardware used to collect photographs and software to

K. McMillan

Department of Geological Sciences, University of

Saskatchewan, 114 Science Place, Saskatoon, SK, Canada

K. McMillan (&)

2501-123 10 Ave. SW, Calgary, AB, Canada T2R 1K8

e-mail: km1138@yahoo.ca

Present Address:

K. McMillan

16 Shawcliffe Bay SW, Calgary, AB, Canada T2Y 1H1

123

J Paleolimnol (2008) 40:1179–1184

DOI 10.1007/s10933-008-9223-5

process the digital images. The hardware consists of

two parts: (1) a rig to which a personal digital camera

is mounted (as on a tripod) and (2) a track for holding

a split core and the camera rig in-line (Fig. 1).

A user manually photographs the core, moving

either the camera rig (‘‘core-stationary mode’’) or the

core (‘‘camera-stationary mode’’) in small increments

(1–2 cm) between photographs (see ‘‘Image collec-

tion: core-stationary versus camera stationary

modes’’ section); the exact spacing between photos

is not important because the final composite image is

created by image-matching software. Shorter inter-

vals between captures yield better results, but longer

intervals use less camera memory.

All dimensions quoted here are from the unit

constructed at the Department of Geological Sci-

ences, University of Saskatchewan, although the

dimensions could easily be modified. Imperial units

are used where construction materials are designed to

such specifications (e.g. 1/4-inch bolts), but otherwise

metric units are given.

The software used in this system is available either

commercially or from public domain. The only nec-

essary processing steps are cropping images and

compositing individual captures into a single core

image. However, some users may also wish to colour

calibrate and/or greyscale calibrate their images for

greater accuracy (see ‘‘Colour and greyscale calibra-

tions’’ section). Several freely available software

packages can be used for image analysis of the final

core image, e.g. Scion Image� (http://www.scioncorp.

com) or ImageJ (http://rsb.info.nih.gov/ij/). Other

freely available programs that users may find helpful

are Coralyzer (http://www.evl.uic.edu/cavern/corewall),

which allows visualization and annotation of core

images, as well as stratigraphic overlaying with other

core data, and PSICAT (http://portal.chronos.org/

gridsphere), which allows users to create multi-data-

set stratigraphic columns from core data.

System components

Track

The track consists of a plywood sheet (1 m long by

47 cm wide) with elevated inner ridges (also ply-

wood, fastened by wood staples) for holding the core

in place, and outer ridges for keeping the camera rig

in-line. A height of 1.5 cm and spacing of 3.5 cm for

the inner ridges is used so that cores with diameters

of about 5–12 cm can easily be accommodated. The

outer ridges are also 1.5 cm high.

Multiple 1-m-long segments of track make storage

easier than a single unit several meters long; four

1-m-long segments can be built from a standard

80 9 40 (1.2 m 9 2.4 m) sheet of plywood. About

12 cm should be left between the end of the core and

the end of the rig so the rig does not roll off the track

during the first or last few captures when operating in

core-stationary mode.

Camera rig

The camera rig is made from standard 200 9 400

lumber and four pre-made wooden wheels (used only

in core-stationary mode). The rig consists of two

trucks bridged by a horizontal crossbar to which a

handheld digital camera is mounted.

Each truck consists of a 20-cm-long horizontal bar

with a 40 cm vertical leg bolted to it. Each horizontal

bar is a piece of 200 9 400 lumber with two 5.5-cm

wheels attached to the outer side by 1/4-inch-

diameter bolts. Each vertical leg consists of two

4.2-cm-wide boards separated by 1/4 inch (to

accommodate the 1/4-inch bolts extending from the

crossbar; see next paragraph); 4.2 cm was chosen for

the board width, to create flush-edges.

The crossbar consists of a 28-cm-long 200 9 400

board with a 1/4-inch-diameter bolt, extending 1/4

inch out of the board, in the middle for mounting the

Fig. 1 Camera rig and track with components as referred to in

text. Upper left inset shows rig disassembled

1180 J Paleolimnol (2008) 40:1179–1184

123

camera (which will have a threaded hole in the

bottom for this purpose), and two 1/4-inch-diameter

end-bolts extending about 4.5 cm from each end for

mounting the crossbar to the vertical legs. The end-

bolts can be fixed in the crossbar by sawing the heads

off four 1/4-inch-diameter bolts and gluing them into

pre-drilled holes in the ends of the crossbar (using

Liquid NailsTM Steel and Metal Framing Adhesive or

a similar wood-metal adhesive) or by fasteners with

screws on one end and bolts on the other. Because

each vertical leg is split along its entire length, the

crossbar can be fixed at any height.

The rig is designed to fit snugly within the track; if

side-to-side movement does occur while rolling,

additional washers can be placed between the wheels

and the trucks to act as spacers.

Scale

All core photographs require an accurate scale for

spatial reference. A length of metal tape measure

works well because it is lightweight and virtually any

length of scale can be created. A nylon tape measure

glued to a flat aluminum dowel also works well, but

should be checked against a ruler after gluing to

ensure it has not been stretched. The scale must be

held at the same height as the split core surface,

otherwise true distances along the core will be greater

than indicated by the scale if it is below the core

surface, and less than indicated if the scale is above

the core surface; the difference in height may also

cause either the scale or the core to appear slightly

out of focus. Figure 2 shows the system in use, with

the scale in place.

Image collection and processing

Image collection: core stationary versus

camera-stationary modes

The system may be operated by moving the core down

the track between successive captures with the camera

fixed in place (camera-stationary mode) or by moving

the camera along the length of the core between

successive captures (core-stationary mode). The main

advantage of camera-stationary mode is that lighting

on the core is constant for each photo, relative to the

camera; thus effectively eliminating the possibility of

uneven illumination of the final core image. The main

advantage of core-stationary mode is that it is more

compact, as the track needs only to be slightly longer

than the core, instead of twice as long.

In core-stationary mode, the scale can be fixed

directly to the core barrel, or propped up next to the

core at the same height as the sediment surface; non-

setting modeller’s clay (PlasticineTM) works well for

propping up the scale; putty (e.g. plumber’s putty,

Silly PuttyTM) is more ductile and may deform under

the weight of the scale.

When operating in camera-stationary mode, the

wheels on the camera rig should be removed or

immobilized with putty, and the scale must be

attached directly to the core barrel. The track should

still be used in camera-stationary mode, because

without it, keeping the core in a straight line between

captures is difficult, and even small deviations may

impede proper photo-compositing. For short cores

(a few metres or less) no rolling mechanism is

necessary to slide the cores along the track. For

longer cores, a series of rollers along the track may

ease operation. Care should be taken to crop all

background out of individual captures, as the back-

ground image will not change systematically with

core position as it does in core-stationary mode; a

discrepancy between core images and background

may also impede proper photo-compositing.

Since either mode may be employed with virtually

no hardware changes to the system, users may choose

to operate in core-stationary mode at some times (e.g.

when photographing core in the field), and switch to

camera-stationary mode at other times.

Fig. 2 Camera rig (seated in track) in use (camera-stationary

mode). Note the scale (mounted with modeller’s clay), and the

digital camera mounted to the cross bar

J Paleolimnol (2008) 40:1179–1184 1181

123

Lighting and camera settings

Different lighting arrangements were experimented

with, and the best results were achieved with

fluorescent room lighting augmented with halogen

spot-lighting. LED lamps were found to be a poor

light source because they do not actually emit white

light, but rather a pale blue light that is poorly suited

to colour photography. Room lighting, used alone,

may be too dim for good photo exposure and may

yield inconsistent shadow effects; however, these

problems can be eliminated by using bright spot-

lights. Care should be taken not to overexpose the

core to spot-lights because the surface of the core

may quickly dry out. Illumination should be as

homogenous as possible to minimize the brightness-

gradient along the core, particularly if using spot-

lights. If necessary, uneven illumination can be

corrected by post-photography processing techniques,

although such corrections may alter real tonal vari-

ations in the image, and may not fully rectify the

heterogeneous illumination (e.g. Nederbragt et al.

2000; Nederbragt and Thurow 2005).

Wet cores often provide the clearest images of

sedimentary structures and fabric; however, surface

glare may prevent proper photo exposure and obscure

parts of the core. The simplest, though not necessarily

most effective way to solve this problem, is to

carefully cover the surface of the cores with trans-

parent plastic film (e.g. kitchen-wrap) (Renberg

1981). Cross-polarization, a more effective technique

for glare-reduction, is discussed elsewhere (e.g.

Lamoureux and Bollmann 2005), but may be difficult

to implement with a personal camera.

The camera should be set to manual exposure, full

colour and maximum sharpness. Many commercial

digital cameras do not offer manual focus, but macro-

focus works reasonably well. The flash should be

turned off. White balance should be set manually by

calibrating the camera with a true white reference

held under the same lighting conditions as the core;

auto-white balance settings may yield inaccurate and

inconsistent tones in photographs, particularly if no

white tones occur in the photo, or if more than one

distinct white tone occurs (e.g. nearly-white sediment

and a white depth-scale).

Exposure needs will vary, depending on the core

being photographed. Cores with alternating dark and

light couplets are particularly difficult to photograph

because parameters must be set so as not to overexpose

the lighter laminae, but also not underexpose the darker

laminae. Users must experiment with different shutter

speeds and aperture diameters (measured in F-numbers

or F-stops) to find an ideal setting for their core.

Unfortunately, most personal digital cameras do not

allow users to set both the shutter speed and F-number

at the same time, and changeable lenses are an unlikely

option, so some trial and error will be necessary to find

the best possible exposure parameters.

Colour and greyscale calibrations

Colour and greyscale calibration is recommended,

particularly if images are to be used for image

analysis. A small colour-checker plate with a set of

known colours and grey-shades, such as the X-Rite/

Gretag-MacbethTM ColorChecker Mini, should be

imaged with every capture to act as the set of

Fig. 3 Schematic diagram

showing relative zones of

distortion of a vertical

downward-facing

photograph

1182 J Paleolimnol (2008) 40:1179–1184

123

reference standards. The plate should be visible in

every photograph so that each captured image can be

colour-calibrated before compositing. The plate

should not be visible in the final composite image,

so care should be taken to place it in a spot that will

be cropped from each individual capture. The actual

calibration can be done using imaging software such

as Scion ImageTM or ImageJ, although add-on macros

may be necessary to perform the task. For more

detailed discussions of colour and greyscale calibra-

tions, see Nederbragt et al. (2005) and Ortiz and

O’Connell (2005).

Image compositing

Vertical photographs of a flat surface are increasingly

distorted as one moves from the principal point

(directly below the centre of the camera lens) to the

edges of the frame (Fig. 3). Because photomontages

require minimally distorted source images, original

images should be cropped so that the most distorted

parts of the photographs are not included in the final

image (Fig. 4). Fortunately, because each photo can

be collected at the same size using a personal digital

camera, the image cropping process can easily be

automated using imaging programs such as ImageJ or

Adobe PhotoshopTM; detailed instructions for using

Adobe Photoshop CSTM are given in Appendix A.

Finally, core images are spliced together using

commercial panorama software. Several packages are

available, and I have found Panorama FactoryTM

(www.panoramafactory.com) to work very well with

a minimum of input commands, although other pro-

grams may work equally well. To minimize possible

image distortion, cylindrical projection should be

used, rather than spherical projection. Partial-

panorama should be selected, rather than full-

panorama, so the last picture in the batch is not

spliced with the first. No automatic image corrections

should be made by the compositing software.

Using the configuration described above, each

metre of core requires about 50–100 individual

photographs. If one collects 100 photos with a 4.1

megapixel camera, about 140 Mb of disk space is

used. Photography takes about 15 min to complete.

After cropping, photo compositing is done automat-

ically by the software, and takes about 5 min to

process (using a 1.86 GHz processor and 1.0 Gb of

RAM). As a reference, the final image is about 6 Mb

in size (at a resolution of about 100 pixels per mm2 of

core), although image size and resolution will depend

Fig. 4 Single original core photograph (shown without colour

calibration plate). Shaded areas are too distorted to use in the

final composite image (note that the ruler seems to move

farther away from the edge of the frame with distance from the

non-shaded part); therefore the image is cropped so that only

the non-shaded area is used

Fig. 5 Final composite

image created from 68

individual photographs

taken in core-stationary

mode (including the one

shown in Fig. 4)

J Paleolimnol (2008) 40:1179–1184 1183

123

on camera height and camera settings, as well as the

image colour-composition. Figure 5 shows an exam-

ple of a finished core image.

Acknowledgements Thanks to Tim Prokopiuk for logistical

support. Reviews by Dan Karasinski and two anonymous

reviewers were very helpful in refining the paper and the

hardware described within.

Appendix A

The following instructions detail how to crop multi-

ple images automatically using Adobe PhotoshopTM

(current to version CS3, December, 2007):

(1) Copy all individual photos from their original

location to a single folder. Keep copies of

original (unedited) images in case of unex-

pected problems during processing

(2) In Photoshop, open any photo to be cropped

(3) Select WINDOW… ACTIONS

(4) Select ‘‘create new action’’ from the small

icons at the bottom of the ACTIONS popup

(5) Give the new action a name, e.g. ‘‘Image

Cropper’’ and click the RECORD icon

(6) Crop the central, least distorted, part of the

photo so that about 1 or 2 cm of core length,

but 100% of the core width, is retained.

NOTE: Some overlap between adjacent

images is necessary for splicing; cropped

images should be slightly longer than the

increment spacing between images.

(7) Click the ‘‘stop recording’’ icon at the bottom

of the ACTIONS popup

(8) Close, but do not save the cropped image

(9) Select FILE… AUTOMATE… BATCH

(10) In the BATCH popup select the following

options:

a. In the ACTION menu select ‘‘Image

Cropper’’

b. In the SOURCE menu select ‘‘folder’’ and

choose the folder with the photos already

in it

c. In the DESTINATION menu, choose

‘‘save and close’’

d. Click OK

Photoshop will automatically open and crop all the

photos in the source folder and save them to the

destination folder. If images are being saved as

JPEGs, Photoshop may prompt the user to verify the

image quality settings each time a file is saved.

References

Cooper MC (1997) The use of digital image analysis in the

study of laminated sediments. J Paleolimnol 19:33–40.

doi:10.1023/A:1007912417389

Francus P, Brandley RS, Thurow T (2005) An introduction to

image analysis, sediments and paleoenvironments. In:

Francus P (ed) Developments in paleoenvironmental

research 7: image analysis, sediments and paleoenviron-

ments. Springer Science and Business Media, Dordrecht,

pp 1–7

Lamoureux SF, Bollmann J (2005) Image acquisition. In:

Francus P (ed) Developments in paleoenvironmental

research 7: image analysis, sediments and paleoenviron-

ments. Springer Science and Business Media, Dordrecht,

pp 11–34

Merrill RB, Beck JW (1995) The ODP color digital imaging

system: color logs of Quaternary sediments from the Santa

Barbara Basin, site 893. In: Kennet JP, Baldauf JG, Lyle

M (eds) Proceedings of the Ocean Drilling Program,

scientific results 146 (part 2), pp 45–59

Nederbragt AJ, Thurow JW (2005) Digital sediment colour

analysis as a method to obtain high resolution climate

proxy records. In: Francus P (ed) Developments in

paleoenvironmental research 7: image analysis, sediments

and paleoenvironments. Springer Science and Business

Media, Dordrecht, pp 105–124

Nederbragt AJ, Thurow JW, Merrill RB (2000) Data report:

colour records from the California Margin: proxy indi-

cators for sediment composition and climate change. In:

Lyle M, Koizumi I, Richter C, Moore TC (eds) Pro-

ceedings of the Ocean Drilling Program, scientific results

167, pp 319–329

Nederbragt AJ, Francus P, Bollmann J (2005) Image calibra-

tion, filtering, and processing. In: Francus P (ed)

Developments in paleoenvironmental research 7: image

analysis, sediments and paleoenvironments. Springer

Science and Business Media, Dordrecht, pp 35–58

Ortiz JD, O’Connell S (2005) Toward a non-linear greyscale

calibration method for legacy photographic collections.

In: Francus P (ed) Developments in paleoenvironmental

research 7: image analysis sediments and paleoenviron-

ments. Springer Science and Business Media, Dordrecht,

pp 125–141

Petterson G, Odgaard BV, Renberg I (1999) Image analysis as

a method to quantify sediment components. J Paleolimnol

22:443–455. doi:10.1023/A:1008070811190

Renberg I (1981) Improved methods for sampling, photo-

graphing and varve-counting of varved lake sediments.

Boreas 10:255–258

Saarinen T, Petterson G (2001) Image analysis techniques. In:

Last WM, Smol JP (eds) Tracking environmental changes

using lake sediments vol. 2: physical and geochemical

methods. Kluwer Academic Publishers, Dordrecht,

pp 23–39

1184 J Paleolimnol (2008) 40:1179–1184

123