It needs to be underlined that documenting the House of Orpheus can be described in general as reality capture
part of the whole project. This means that its main goal was to collect the maximum available spatial information
about the site that will enable its further post-processing with best possible accuracy in the office conditions.
The core element of that process is undoubtedly the metric survey which delivers the 2 and, in this case,
3 dimensional datasets being the mathematical description of site’s spatial arrangement, division and structural elements.
This information can be further transformed into various documentation level like 3D models
and their 2D representations - orthoimages or drawings.
Nowadays among the methods most efficient for making an accurate, fast and reliable metric survey the remote sensing seems to be proper and only choice.
Within that toolbox laser scanning and photogrammetry deliver possibly best solutions in modern surveying.
For this reason both of the methods were chosen for capturing House of Orpheus. On the one hand laser scanning
that provides surveyor with most accurate information about geometry of documented surfaces on the other
photogrammetry that enhances their visual layer. Every reality project is in its nature a compromise between
the expected accuracy, delivery of post- processed documentation, storage capacity and computing power of surveyor
and end-user computers. That fact is directly related to site’s dimensions and the level of detail that needs to be detected and visible.
Simply it is not necessary to capture the whole city with mm quality to analyse it but uncovered finds need to be measured with even better accuracy.
The House of Orpheus project is in that case something in between as it represents a piece of residential architectural
structures within the city with plenty of stone elements that required greater accuracy.
Taking into consideration presented facts and also site’s dimensions and volume - represented in general
as a maximum depth of lowest deposits below current terrain level, the 4mm quality for architectural structures
located in situ seemed to be the most reasonable choice. Achieving that level of detail required making of 62
terrestrial laser scans with 28M points resolution. During fieldwork scanning process followed the grid
of uncovered rooms visible on site. As it is seen from a simple calculation the number of points registered
in raw point clouds reached 1,8B in total. Raw point clouds - basic scanning datasets were later on filtered,
colour corrected, registered and referenced. However 1,8B points is clearly a number impossible to handle
not only by the end user but also difficult to process further especially together with photogrammetry data.
So finally the main data cluster was reduced to step 4 and 450M points unified point cloud was used for post- production.
Simultaneously the photogrammetric measurements were carried on to ensure the best global dataset consistency.
In total 1300+ photogrammetry images were captured. Among them 900 came from extensive aerial UAV survey, the rest were done with classic,
close range, terrestrial methods. The images were taken with full frame 24Mpix and 21Mpix 1’’ sensor camera.
If we now compare the pixel to point number it becomes clear the number of image data is almost the same as
scanning points and reaches 1,6 Gpix. Photogrammetry pictures were colour corrected, oriented, referenced and processed
to raw photogrammetry surface model built of 150M vertices. At the end the 450M point cloud model and 150M vertex model
were blended together and final 3D surface model was created where the quality of geometry came from 4mm laser scans
but the texture - RGB colour from photogrammetry pictures. Presented model became a foundation for processing the rest
of documentation. It was cut into orthogonally projected slices to create all necessary sections, plans or other rendered
documentation or presentational materials. The referenced orthoimagery was further imported into GIS environment to create
analytical vector documentation and databases. It is worth concluding that in fact only 25% of raw data was finally used
for the metric survey, so even more accurate than 4mm solutions will be available in the future.
As mentioned at the beginning the level of detail proper for capturing architecture may not fit exactly the accuracy
suitable for smaller artefacts. For the reason for capturing the selected finds, mostly stone, architectural details
other remote sensing equipment an strategy must have been implemented. In that case the SLS scanner was used for documenting
the series of archaeological finds with sub millimetre - 0,5 to 0,7mm quality. Those scans were directly processed to 3D textured models
in the formats common for 3rd party 3D modelling applications. The usage of phase based scanning for architecture,
SLS for details combined with photogrammetry resulted the creation of complete metric survey, core dataset
for both 2D and 3D analytical purposes. Among all, high quality and detailed 3D models could have become a real world
reference for further digital reconstruction of the House of Orpheus based on accurately registered site context.