the timber 
es are gone, 
>cted areas, 
torical city 
\iums, cars, 
O create, as 
limensional 
ent City of 
n animated 
1e 300 years 
ural models 
ons, models 
le, models of 
f this can be 
chnique of 
e existing 
1 sources are 
| will also 
traditional 
typical to the 
main of the 
ic recording 
ential part of 
.ogrammetry 
desired end 
ge. 
of existing 
togrammetry 
reports as to 
the missing 
f the missing 
ttributes of 
lephants, the 
use in video 
into a video- 
nputer based 
?CORDING 
nstruction of 
ig point from 
of the well 
: recording 
netry (Ogleby 
  
1995,) it was an obvious choice as a tool to create 
the initial models of the main remaining 
architectural elements. In addition, because of 
the initial lack of a suitable base map, small 
format aerial photography was also planned as 
the method offered the possibility for creating 
three-dimensional maps to a sufficient accuracy 
(Fraser et al, 1995). 
4. Architectural Recording 
The technique employed at  Ayutthaya is 
relatively straight forward, and based on (but 
notlimited to) the proven case of the terrestrial 
stereo-pair. In general, stereo-pairs were taken 
of the facades of selected chedis, prangs, stele 
and walls so that a ‘library’ of architectural 
elements could be derived. Convergent 
photographs were also taken of many of the 
features so as to have the possibility of 
strengthening the control network for the 
features if necessary. The camera used was a 
Hasselblad 500ELM with a reséau plate, and 
either a 50mm lens or an 80mm lens as required 
by the geometry. The camera was mounted on a 
good quality, stable, Manfrotto brand tripod, 
allowing the camera orientation to be repeated 
between exposure stations to an accuracy 
suitable for the restitution of the stereo-pairs. 
The film used was Fujichrome 100 ASA colour 
transparency, giving good colour saturation 
with medium archival qualities. 
  
Figure 2: Wat Kung Sang, a chedi in poor condition 
It is important to acknowledge that the 
photogrammetric recording procedure was 
designed for the purpose of creating the basic 
three dimensional models needed for the 
419 
computer reconstruction. This was not a 
documentation program, although the records 
can be used in part for this purpose. Rarely were 
all the sides of a chedi recorded, a minimum 
number of elevations or facades were 
photographed as nearly all of the architectural 
features were symmetrical. Poor or badly 
preserved examples (Figure 2) of a particular 
chedi or prang style were not recorded, the 
photographs would not be of much use in the 
derivation of a three dimensional ‘cell’ of the 
object. There was neither the time, the funding 
nor the immediate level of interest in 
establishing a significant documentation 
program as part of this project. 
Numbered targets were placed on the surface of 
the features being photographed, and 
coordinates derived for each of these. The 
targets are made from 5cm square aluminium 
covered with a self-adhesive cross target 
produced some years back for a theodolite 
intersection exercise in the aerospace industry. 
The targets are numbered and placed on the 
feature using re-useable adhesive similar to the 
‘BluTac’ product, allowing the targets to be 
removed once the photography is completed. In 
general, six such targets are placed for each 
stereo-pair as experience has shown the extra 
time taken to coordinate targets in the field is 
considerably less than that required to extend 
weak control networks across architectural 
features using either triangulation or bundle 
adjustment of pass points. Where the size or 
height of the object did not facilitate the 
placement of targets at key locations, features on 
the surface of the building were used as photo- 
control points. Their location was described and 
sketched so they could be recognised back in 
Australia. 
Three dimensional coordinates for the targets 
was provided by theodolite radiation from either 
an arbitrary control point, or an instrument 
point coordinated as part of a network. A 
network of controlled instrument points were 
placed around the main chedi area of Wat Pra Si 
Sampet (the most recognisable remains at 
Ayutthaya), and coordinated onto the Thai 
mapping grid using a Leica System 300 GPS as 
part of the preparation for the aerial 
photography. This provides for the actual 
spatial location of the eight features recorded to 
be determined. At other locations a local 
coordinate datum was used as only the shape and 
size of the feature was required, its true position 
could be determined by other means if necessary. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996