Multi-Scale and Multi-Sensor 3D Documentation of Heritage Complexes in Urban Areas^†

Round 1

Reviewer 1 Report

The paper is well structured. Some clarifications/suggestions as below can improve its quality:

Lines 210-215: Please explain and elaborate why there are problems in measuring distances, while there were no errors in distances. Typically, total stations are designed to work with no problems from -50oC to +50oC

Lines 246-249: Please elaborate on the small number of GCPs (5) and of Check Points (3). Such small numbers prevent a sound statistical inference on accuracy and reliability. Please explain.

Line 305: Please explain the meaning of “….TLS data was georeferenced and lacks a scaling factor,…”

Line 321: Please explain the meaning of “…This further shows one of the disadvantages of photogrammetry, namely the noise generated by movable objects…”. In, fact, what you map is the immovable buildings. So, any “noise” in movable objects does not actually counts.

Lines 370-371: Please explain “…This might be caused by the sensor to object distance, as both photogrammetry and TLS data were acquired from a slightly farther distance than that of the plinth…”

Lines 468-469: Besides your comment “…The TLS used in the project simply is not designed for such close range application…”, it should be mentioned that there are Laser scanners (other models than the one used here) that are suitable for close-range applications (eg. Minolta)

Lines 473-474: Please explain what are these “robust metrics” are in “…The analysis in this paper also showed that more robust metrics are sometimes required in order to better represent…”

English language and grammatical error corrections in various places are needed.

Author Response

The authors would like to thank the reviewers for their time and expertise invested in this review. Some of the changes to the paper have been marked in yellow in the annotated document. We will try to comment on the reviews point by point.

The paper is well structured. Some clarifications/suggestions as below can improve its quality:

Lines 210-215: Please explain and elaborate why there are problems in measuring distances, while there were no errors in distances. Typically, total stations are designed to work with no problems from -50oC to +50oC

We thank the reviewer for this interesting remark. Indeed, while total stations are theoretically capable of operating under extreme temperatures, we have encountered this particular problem several times (not only during this project), mainly with the distance meter which utilises infrared wave. The problem occurs mainly when using the prism-less measurement mode (for detail points). Although it is not always the case, sometimes either the measurement is long and/or difficult, or the measured distance is wrong. The angular measurement showed no problem at all. In hindsight, the strong sun exposure may have also played a role in this regard. In order to minimise potential error from this, we decided to perform spatial intersection which also has the added advantage of being more robust (being measured from at least two stations). A further description of the problem has been added to the paper.

Lines 246-249: Please elaborate on the small number of GCPs (5) and of Check Points (3). Such small numbers prevent a sound statistical inference on accuracy and reliability. Please explain.

Thank you for this comment. We agree that the number of control and check points is rather inadequate to give sound statistics. This is actually hampered by project constraints. Since the main objective of the step 1 processing was to give a general overview of the site, not enough time was available to measure too many control points for the whole area (the main area of interest for the project being the Siti Inggil area). While far from ideal, this configuration was largely enough for our step 1 requirements. A similar explanation has been added to the paper.

Line 305: Please explain the meaning of “….TLS data was georeferenced and lacks a scaling factor,…”

In this phrase, we want to express one of the advantages of TLS with regards to photogrammetry, mainly that the TLS does not require scaling factors as it is by nature a repetition of distance measurements. Furthermore, the TLS data in our case was georeferenced by the measured coordinates of some artificial spheres. It is for this reason that we think that the TLS data may be used to georeference the close range photogrammetry models in view of the lack of control points. A similar explanation has been added to the paper.

Line 321: Please explain the meaning of “…This further shows one of the disadvantages of photogrammetry, namely the noise generated by movable objects…”. In, fact, what you map is the immovable buildings. So, any “noise” in movable objects does not actually counts.

Thank you for this comment. We agree that the noise actually comes from objects that we do not necessarily map per se. We have decided to delete the phrase.

Lines 370-371: Please explain “…This might be caused by the sensor to object distance, as both photogrammetry and TLS data were acquired from a slightly farther distance than that of the plinth…”

Thank you for this remark. In the case of the ceiling, the sensor to object distance was farther than that of the plinth, which reduces the resolution of both photogrammetry and TLS data as a result. This may have played a role in the fact that the standard deviation for the ceiling is slightly higher than that for the plinth. More detailed description of this reasoning has been added to the text.

Lines 468-469: Besides your comment “…The TLS used in the project simply is not designed for such close range application…”, it should be mentioned that there are Laser scanners (other models than the one used here) that are suitable for close-range applications (eg. Minolta)

Thank you for this important input. We agree that this is a very important information to put in the paper. The appropriate examples of such close range laser scanners have been added to the text.

Lines 473-474: Please explain what are these “robust metrics” are in “…The analysis in this paper also showed that more robust metrics are sometimes required in order to better represent…”

Thank you for this remark. We agree that the word robust may be misplaced in this regard. We have therefore modified it to reflect this change.

English language and grammatical error corrections in various places are needed.

Thank you for this input. A more thorough proofreading of the manuscript was performed to address this problem.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper is well written and describes a good approach and extensive work flow from data capture to processing to modelling. 

Because of the extent of the work details are sparse on processing of data. 

It is clear how the point cloud and photogrammetry were analyzed in cloud compare but more details should be included regards the process for combining the laser and photogrammetry data to create a United Model. 

The HBIM element is sketchy and creating the model from the vectorised mesh in ArchiCAD needs more detail how was the texture brought into HBIM. 

Would it not be easier to use imported meshes for parts of the model where you have complex geometric forms. 

Overall the paper is of interest to researchers in the field as it described a complete workflow for recording documenting and presenting Architectural Heritage. 

Author Response

The authors would like to thank the reviewers for their time and expertise invested in this review. Some of the changes to the paper have been marked in yellow in the annotated document. We will try to comment on the reviews point by point.

This paper is well written and describes a good approach and extensive work flow from data capture to processing to modelling. 

Because of the extent of the work details are sparse on processing of data. 

It is clear how the point cloud and photogrammetry were analyzed in cloud compare but more details should be included regards the process for combining the laser and photogrammetry data to create a United Model. 

Thank you for this remark. We agree that the combining of TLS and photogrammetry is a very important part of the paper. We have added more details in sections 3 and specifically 5.3.2 in the case of the use of the united model in creating a photo-realist meshed model.

The HBIM element is sketchy and creating the model from the vectorised mesh in ArchiCAD needs more detail how was the texture brought into HBIM. 

Would it not be easier to use imported meshes for parts of the model where you have complex geometric forms. 

Thank you for this input. Indeed, the HBIM part was a test and was not the main aim in this stage of the project. For the moment the texture in the HBIM model is not the photo textures, but rather an artificial repeating pattern texture image. Indeed, the combination of mesh and geometric primitives is an interesting idea and may be addressed in a further work. A similar explanation has been added to the text to clarify the nature of the HBIM model.

Overall the paper is of interest to researchers in the field as it described a complete workflow for recording documenting and presenting Architectural Heritage. 

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper present a method to survey and document a cultural heritage site. Even if the methodology of integrating multi-scale surveys is not new, the process of the work, the accuracy of the results and the review of the problems and solutions are appropriate and incisive.

Comments for author File: Comments.pdf

Author Response

The authors would like to thank the reviewers for their time and expertise invested in this review. Some of the changes to the paper have been marked in yellow in the annotated document. We will try to comment on the reviews point by point.

The paper present a method to survey and document a cultural heritage site. Even if the methodology of integrating multi-scale surveys is not new, the process of the work, the accuracy of the results and the review of the problems and solutions are appropriate and incisive. 

Lines 246 – 257: You say that 8 control points were measured, supposedly georeferenced via GNSS. Why then is the use of TLS data as reference for the orientation needed? Can you explain?

We thank the reviewer for this question. Indeed, the 8 control points mentioned in lines 246-257 refer to aerial control points used in Step 1 of our project division. These points are scattered over the whole area of the palace, and is therefore insufficient to georeference Steps 2, 3 and 4. Indeed, a topographical survey was conducted in parallel with the TLS acquisition which enables it to be georeferenced. This is afterwards used to georeference larger-scale objects obtained using close range photogrammetry. The use of this method (deriving close range photogrammetric control points from georeferenced TLS data) enabled us to reduce significantly the time required to perform the topographical survey, which often times present the longest part of a field acquisition. Nevertheless, we do agree that in the case of lines 246-257, the distinction between Step 1 GCP and the points measured for the TLS georeferencing during the topographical survey was not clear. We have modified the text to better reflect this distinction.

Lines 310 and following: the movable objects interfered in the comparison of the different data. Why didn’t you think about deleting for example the trees from the different point clouds before comparing them in order to lower the possibility of deviation between the different data?

Thank you for this remark. We agree that the movable objects interfered in this case, and that a more sound comparison should be performed by deleting these moveable elements. Initially, we thought about giving an “as is” comparison, amongst others to also show how photogrammetry fares in modeling the tree crowns. This has been added to the text to better show the context.

Figure 7: the comparison seems to show a slight rotation of the two models combined. This seems to be derived from a not accurate registration. Have you thought about redoing the alignment in order to have a better registration? In case you tried and the results were always the same, how have you explained the result?

Indeed, we also noticed the effect of the slight rotation of the two models. We think that it is due to the lack of points for the photogrammetric data at the level of the floor tiles and also at the back of the object, which is heavily covered by vegetation. We think that this plays a role during the ICP process, which resulted in the slight rotation as observed in Figure 7. This additional explanation has been added to the text.

Paragraph 5.3.3: For the BIM, have you used profiles derived from the 3D models to create it?

Thank you for this question. For the creation of the BIM, profiles were derived from the photogrammetric (wireframe with surfaces) 3D model generated in the software PhotoModeler.

Paragraph 5.3.4: Have you thought about testing the VR applications with public?

We do have experience in testing VR applications with the public in previous projects. In most cases, the results are overwhelmingly satisfactory to the public as well as the stakeholders, in that it manages to convince the public about the importance of heritage documentation. In this particular project described in the paper, plans were made to do the same public demonstration of the VR system to Cirebon city officials and representatives of the palace. It was, however, hindered by logistical problems linked to the necessity to transport the required items (VR goggles, computers, etc.) from France to Indonesia. Another mission to Indonesia in 2019 is planned to better tackle this logistical problem and show the VR of the Kasepuhan Palace to the Indonesian public.

An overall check of the English is needed.

Thank you for this input. A more thorough proofreading of the manuscript was performed to address this problem.

Author Response File: Author Response.pdf