CIPA 2005 XX International Symposium. 26 September - 01 October. 2005, Torino, Italy, 
632 
2. ARCHITECTURAL DOCUMENTATION 
2.1. Architectural documentation targets 
There are several reasons to manage an architectural 
documentation project. 
2.1.1 Conservation: An International authority or organization, 
declare that a build work belongs to universal cultural heritage. 
Its physical conditions has to be managed properly so that it can 
be preserved and prove its cultural specificity. Building 
restoration provoke the modification of initial state. The 
documentation of an architectural work in such a condition 
implies its complete understanding (how it had been build, its 
original occupation, for example). In a case like this, there is a 
need for precise dimensional dataset and appropriate 
management of it. 
2.1.2 Promotion: Cultural heritage is a part of the urban 
identity and the community can thus promote itself by 
emphasizing it (Figure 1). In such a case the documentation is 
focussed on representation and communication. 
Figure 1. Water Tower (VRML and Photograph) 
2.2. Three-dimensional data acquisition techniques 
In the field of 3D data acquisition we can find two kinds of 
techniques: Scanning and Photogrammetry. They represent two 
opposite approaches: active and passive mesure. We can 
associate this antagonism with the difference between the 
automation ability provided by scanning techniques and the 
difficulty to obtain architectural data without human 
intervention considering Photogrammetry (more and more 
reduced but still significant). 
2.2.1 Scanning: On one hand lay LASER scanners and their 
vast application field (Fuchs, 2004). On the other hand another 
technology is being developed, a technique based on structured 
light theories, which promises lot’s of advantageous uses in 
architectural detail survey. This method has the property to be 
reachable with cheap tools (Batlle, 1998). Those automatically 
produce point clouds. Post-processing steps are necessary to 
integrate this kind of data in an architectural modelling process. 
2.2.2 Photogrammetry: 3D acquisition remote techniques 
originate from this discipline. There are many different 
approaches to handle architectural survey with Photogrammetry 
(Grussenmeyer, 2002; Saint-Aubin, 1992). A great advantage of 
using photographs as measurement source is the fact that they 
allow to carry out photo-interpretations, which are helpful in the 
modelling process. At the end of restitution step, we can obtain 
2D plan or 3D vectors. Thanks to correlation, it is possible to 
get point cloud automatically but only when the project has 
already been oriented 
2.3. Encountered problems 
Even if dimensional data are numerous and very precise, we 
always miss a part of the build work, primarily due to the mask 
phenomenon. Masks are hiding a part of the object to be 
surveyed and create a "hole" in the collected dataset (Alby, 
2004 & Figure 2). Missing part of build work can also provoke 
such a deficiency of data. As the buildings are the result of 
architectural design, there are parts which have been repeated. 
Consequently the acquired data contains duplication of the same 
observable fact. There is therefore a discussion to manage about 
the significance of the architectural model we want to carry out 
based on such dataset and the knowledge to bring during the 
modelling process. 
Figure 2. Mask due to projection and building itself 
3. GEOMETRICAL MODEL AND DATA 
3.1. Geometry 
In this part, the differentiation between 3D data and geometry 
will be highlighted. 
3.1.1. Model: Scientifically, modelling is a reconstruction 
process of the studied object in order to manipulate it. In 
architectural documentation, 3D data are often directly 
translated into representation, due to the fact that surveyor and 
architects (or archaeologist) are distinct person. But it would be 
interesting, for two reasons, that providers and users focus on 
the model: firstly the increasing use of computerised data 
induces the user to need other complex information instead of 
representation; secondly, during representation step, a provider 
has to make hypothesis which does not necessarily belong to his 
attribution. The model is a geometrical model. In this case 
geometry is good enough to describe architectural works for 
restitution of shapes and dimensional query. 
3.1.2. Acquisition: Data collected during an acquisition 
operation has to be related directly with the real object. But this 
is just a mean to construct its 3D model. Based on the preceding 
paragraph, the 3D data can be seen as a relation between the 
model and the reality but not the only source of definition of 
geometry. Topology is another method for the preliminary 
definition of the model. It is very important to consider 3D data 
as one tool among others in a first approach. Even if it could 
have a central position in the provider point of view. It is 
necessary to separate, intellectually, the field of acquisition 
from architectural documentation, to reconnect it later but 
differently.