The Application of Photogrammetric Survey ... the Great Ceiling at Peterborough Cathedral 
95 
floor, the camera and tripod were 
mounted on a dolly, which allowed the 
whole assembly to be simply wheeled 
from one camera location to another. 
Given the symmetrical coverage offered 
by the UMK, the camera was wheeled 
straight down the centre of the nave with 
each photograph being taken vertically 
upwards under each of the central 
lozenges. 
This resulted in a total of 23 photo 
graphs being taken on both colour and 
black and white negative film, during 
November 1996, providing 22 stereo 
models for later processing. In general 
this arrangement worked very well 
although the presence of the large 
Crucifix, hanging down from the ceiling 
itself, did cause an obstruction and 
hence required more stereo-photography 
than originally anticipated. 
It was always expected that due to the 
sheer size of the ceiling the correct 
exposure of these ground based images, 
particularly in colour, would prove to be 
a problem. The combination of the 
height of the ceiling and the difficulty of 
safely placing a powerful enough fash 
system within the cathedral, to 
illuminate the full width of the ceiling in one exposure, meant that the only feasible option was to use the available natural light. This 
appeared to be relatively consistent, during overcast conditions, although the effect of light sometimes cascading in through the 
clerestory windows at upper level, meant that photography had to be taken during late afternoon to reduce the likelihood of fare. On 
the day of the photography an exposure of Is at f22, onto Kodak Vericolour colour negative film rated at 160 ASA, seemed to be 
sufficient although the lighting regime used was to prove a major problem during later processing, particularly in successfully 
managing the overall colour balance of the ceiling images. 
Figure 4: Stereo pair taken with Zeiss UMK 30/1318 metric camera of eastern most 
bay 
In addition to this ground based photography a series of stereo-images were also taken from the clerestory level using the smaller 
5"x4" (13 x 10 cm) format WILD P31 metric camera. These were taken principally to cover the two canted side sections of the 
ceiling and the vertical ashlar boarding, but to date these have not been required due to the excellent coverage provided by the 23 
ground based UMK images. 
3. SURVEY CONTROL 
To enable each of the 22 stereo models to be satisfactorily orientated in either a traditional analytical plotter or a digital 
photogrammetric workstation, a minimum of 4 control points per model are normally required. Typically these are applied butterfly- 
style targets, allowing angle intersections from a fixed baseline to be accurately observed. For the ceiling however the safe placement 
and subsequent removal of around 90 of these targets was deemed impracticable. Therefore it was decided to use points of detail, 
even though it was accepted that these would not provide the same clarity of pointing and hence level of accuracy as a fixed target 
and would also take a lot longer to observe. To allow these intersections to be carried out effectively a set of colour prints were made 
from the stereo-photography which were marked up directly on site, as detail was selected and co-ordinated. The wide platform at 
clerestory level, on both sides of the aisle, provided the best location for survey observations, giving an excellent view of the ceiling 
as well as reducing the need for any extreme vertical angle observations to be taken to the points of detail. 
At the time of the survey, December 1996, there was no fixed survey grid existing around the cathedral into which these observations 
could be tied. Therefore a total of 20 temporary survey stations were set out on a local grid at clerestory level within the nave of the 
Cathedral, from which intersection observations to 87 points of detail were taken on both faces using a LEICA TCA 1 103 total 
station theodolite. The motorised horizontal and vertical circles and onboard PCMCIA data storage card ot this theodolite greatly 
facilitated the observation work. The observations were all computed off site using the Landscape processing package to provide the 
required 3D co-ordinates. This processing provided 85 detail points where the height residual between observations was less than 
5mm, and only two where the error was 6mm. Due to the anticipated problems in observing to points of detail instead of fixed 
targets, these observations were deemed acceptable for the processing work anticipated. Also the arrangement selected on site 
provided at least 6 control points per stereo-image and hence a greater degree of redundancy during the orientation processes to 
come.