TouchTerrain—3D Printable Terrain Models
Round 1
Reviewer 1 Report
My views on this submission are rather mixed. Overall, it is very interesting and does make a useful contribution to the discussion on multi-media, so does have a place in the special issue. However, much of it is descriptive rather than analytical, and much of the discussion of the use of models produced is anecdotal (as is acknowledged) rather than reporting on a structured assessment of use and users. Some aspects could be considered purely advertising the system, rather than contributing to scientific debate.
It is good to see a relatively wide-ranging literature review which is often lacking. It is a shame that the review of 3D model use is not built on more thoroughly in the latter part of the paper.
The long bullet point list of facilities added to this version is largely promotional rather than adding to scientific debate. Users should be able to access details of the capabilities from the website or manuals, not a scientific journal.
In addition to the about there are also some stylistic issues which need to be addressed. It is rather long and the writing could be more focused, It would not be reasonable to expect a total re-write at this stage, but some significant editing would make it more appropriate for the journal.
Specific observations:
P2, L58 – incorrect bracket
P2 and subsequently – there is an inconsistency in referencing. Mostly citations are by number in brackets, but in some places by author & year. One system should be adopted throughout.
P3, l96 ‘… that can run out.’
P3 l127 – full stop which should be comma
P4 l164. Starting a paragraph or sentence is this way is unattractive. In this case it would be better to use the author’s name first, followed by the numerical reference, or re-write the sentence to place the reference later. This occurs repeatedly …
P4 l166 UK Ordnance Survey does not exist. It should be ‘Ordnance Survey of Great Britain’, or perhaps at that time the maps used would be titled ‘Ordnance Survey of England’.
P4 l180 see earlier comment about paragraph leading with a citation
P5 l234 – poor style of referencing. ‘A study performed by [15] …’ 15 refers to a publication, not just the authors. At least ‘The study reported in [15], but even this is lazy use of grammar.
P5 l243 style again
P6 – several style issues
P6 l 271 – I think tactile maps with raised symbols is quite different to representing the terrain in 3D – there is a lot published over the years about tactile maps.
P13, l584 and onward. The process of corner interpolation seems convoluted. The DTM values can directly be considered values of corners of squares as is done in contour threading or plotting a 3D mesh block diagram, where in effect we have a series of profiles in x & y running through the DEM values. Viewing the values as cell centres or corners is effectively an offset of ½ the grid dimension in terms of the area of the ‘map’. Directly using the DEM values would reduce processing, eliminate the need to create fictitious values, and remove the needless additional generalisation of the data that must occur through further interpolation.
P14 figure 5. No A-D labels. I really don’t understand why the positions of corners are not in a regular grid arrangement in the top right figure, if this is truly viewed from directly above. Is this a perspective view from close to the 3D model surface, rather than an orthogonal view?
P16 & onward – Sections 3.2.3, 4 & 5. Some of this reads a bit like an abbreviated manual for the software. This should be reduced to the bare minimum
P19 I think it would be sufficient to say the bottom relief can be in Braille, without figure 11.
Figs 12 & 13. It would be preferable if both of these maps were global, and certainly of the same extents.
P24 fig 15. A much more sensible horizontal axis labelling could be used in the upper figure. We seem to have an interval of 316.9 sq km! In fact, I assume each column is a range, so why not an interval of 500 sq km which would greatly simplify the labels?
In the lower plot, the horizontal axis values are meaningless. Ok, it is on a log scale, but the axis could be labelled with sq km values (probably rounded to the nearest sq km).
P24, l893. It would actually be interesting to know how many were ‘just testing the software’ – which is a positive response compared to gibberish. You quote 6% of users provided comments the give the number of valid ones as an absolute figure, which makes comparison difficult – add the total number of respondents, or the number discarded.
P27, table 3. While interesting, it is a shame that the openness of the categories – covering both activities (eg education) or product (map), and how they potentially overlap makes any meaningful analysis of sub-categories impossible.
P30, l995 – Author/date citation out of keeping with other citations.
Appendix A is of little value without further information, such as the size (or extent) of image, the location of the user, if the 3D model is actually created, etc. Some summary of part of the world or type of feature mapped would at least add some analysis. As it stands, it could easily be reduced to the top 10 or top 25 as something of general interest.
Although there is a list of student comments in appendix B, the discussion of them in the text is very limited and there is no attempt an analysis.
Appendix C is largely advertising the programme, a couple of examples would suffice to illustrate the capabilities. Hand painting is very much an additional effort and not part of the model production process.
Appendix D is interesting and has some useful guidance, but is not well written and should be considered more effectively in the main body of the text. I think a ‘Best practice’ guide needs to be a bit more comprehensive, especially about issues such as suitable terrains, vertical scaling, size of model, etc.
Author Response
Reviewer 1 responses (in bold):
My views on this submission are rather mixed. Overall, it is very interesting and does make a useful contribution to the discussion on multi-media, so does have a place in the special issue. However, much of it is descriptive rather than analytical, and much of the discussion of the use of models produced is anecdotal (as is acknowledged) rather than reporting on a structured assessment of use and users. Some aspects could be considered purely advertising the system, rather than contributing to scientific debate.
It is good to see a relatively wide-ranging literature review which is often lacking. It is a shame that the review of 3D model use is not built on more thoroughly in the latter part of the paper.
The long bullet point list of facilities added to this version is largely promotional rather than adding to scientific debate. Users should be able to access details of the capabilities from the website or manuals, not a scientific journal. The list of TouchTerrain version 3 improvements was removed.
In addition to the about there are also some stylistic issues which need to be addressed. It is rather long and the writing could be more focused, It would not be reasonable to expect a total re-write at this stage, but some significant editing would make it more appropriate for the journal. An attempt was made to tighten up the writing and to remove superfluous information. Despite adding ~ 2 pages worth of new text in response to reviewer comments, the total number of pages dropped from 40 to 32.
Specific observations:
P2, L58 – incorrect bracket Fixed
P2 and subsequently – there is an inconsistency in referencing. Mostly citations are by number in brackets, but in some places by author & year. One system should be adopted throughout. Author citations were a mistake, number citations are now used (except when leading a paragraph, where author and number are used)
P3, l96 ‘… that can run out.’ Fixed
P3 l127 – full stop which should be comma Fixed
P4 l164. Starting a paragraph or sentence is this way is unattractive. In this case it would be better to use the author’s name first, followed by the numerical reference, or re-write the sentence to place the reference later. This occurs repeatedly … In these cases the name of the author(s), followed by the regular citation number are now used.
P4 l166 UK Ordnance Survey does not exist. It should be ‘Ordnance Survey of Great Britain’, or perhaps at that time the maps used would be titled ‘Ordnance Survey of England’. Changed to Ordnance Survey of England
P4 l180 see earlier comment about paragraph leading with a citation. Fixed
P5 l234 – poor style of referencing. ‘A study performed by [15] …’ 15 refers to a publication, not just the authors. At least ‘The study reported in [15], but even this is lazy use of grammar. Added the name of authors
P5 l243 style again Added the name of authors
P6 – several style issues
P6 l 271 – I think tactile maps with raised symbols is quite different to representing the terrain in 3D – there is a lot published over the years about tactile maps. Clarified that traditional tactile (paper) maps use raised symbols and added a reference. Further clarified that these 3D printed tactile maps are based on 3D terrain with added tactile symbology.
P13, l584 and onward. The process of corner interpolation seems convoluted. The DTM values can directly be considered values of corners of squares as is done in contour threading or plotting a 3D mesh block diagram, where in effect we have a series of profiles in x & y running through the DEM values. Viewing the values as cell centres or corners is effectively an offset of ½ the grid dimension in terms of the area of the ‘map’. Directly using the DEM values would reduce processing, eliminate the need to create fictitious values, and remove the needless additional generalisation of the data that must occur through further interpolation. Agree that the process appears convoluted as described and should be re-implemented more elegantly as a half offset. The current way does however afford several positive aspects with regard to multi tile management and undefined (empty) cell values, but these would be even more laborious to describe in detail. As the paper is already way too long and the concepts were already described in our 2017 paper, these details are now omitted and the entire section has been shortened.
P14 figure 5. No A-D labels. I really don’t understand why the positions of corners are not in a regular grid arrangement in the top right figure, if this is truly viewed from directly above. Is this a perspective view from close to the 3D model surface, rather than an orthogonal view? Added labels. Yes, the visualization software we used uses a perspective view instead of a pure orthogonal view. The view is now slightly tilted and better shows the corner box graphics to communicate the upwards extrusion of the cell centers.
P16 & onward – Sections 3.2.3, 4 & 5. Some of this reads a bit like an abbreviated manual for the software. This should be reduced to the bare minimum. Sections from 3.2.3 up to section 4.0 were reduced
P19 I think it would be sufficient to say the bottom relief can be in Braille, without figure 11. Done
Figs 12 & 13. It would be preferable if both of these maps were global, and certainly of the same extents. Done
P24 fig 15. A much more sensible horizontal axis labelling could be used in the upper figure. We seem to have an interval of 316.9 sq km! In fact, I assume each column is a range, so why not an interval of 500 sq km which would greatly simplify the labels? In the lower plot, the horizontal axis values are meaningless. Ok, it is on a log scale, but the axis could be labelled with sq km values (probably rounded to the nearest sq km). These 2 plots were replaced by a single log 10 plot that shows km2
P24, l893. It would actually be interesting to know how many were ‘just testing the software’ – which is a positive response compared to gibberish. You quote 6% of users provided comments then give the number of valid ones as an absolute figure, which makes comparison difficult – add the total number of respondents, or the number discarded. Added numbers to all of them.
P27, table 3. While interesting, it is a shame that the openness of the categories – covering both activities (eg education) or product (map), and how they potentially overlap makes any meaningful analysis of sub-categories impossible. Due to the wide variability and large number of feedback texts, it was difficult to design a stringent system based on e.g. only activities. This does limit its analytical use but is hopefully still informative. We attempted to extract the “best” subcategory. For example a text might mention mountains, so it was classified as Map – Mountains. Another might also mention a mountain but in a national park, this was then classified as Map – Park b/c it was seen as more narrow. Another text might point to a national part with no mountains, which was also classified as Map – Park. Ideally we would set up some sort of ontology to capture such subtleties but this was deemed overkill for this paper. However, to improve understanding of the nature of the categories and sub categories we’ve includes example text for the major combinations (>5 hits)
P30, l995 – Author/date citation out of keeping with other citations. Fixed
Appendix A is of little value without further information, such as the size (or extent) of image, the location of the user, if the 3D model is actually created, etc. Some summary of part of the world or type of feature mapped would at least add some analysis. As it stands, it could easily be reduced to the top 10 or top 25 as something of general interest. Note that Appendix A only shows search results. These results are strictly speaking not connected to the areas users actually downloaded later. Granted, after a search with Grand Canyon as term, the user would very likely eventually download and 3D print a model of the Grand Canyon. However, our current system is not set up to actually link the search result to the area for which the download event was later triggered, i.e. the user could certainly have moved the map manually to another area on the globe and download that area instead. As it is, it is just an indication of what areas users were initially interested in, hence it’s relegations to the appendix. This was clarified and the number was reduced to Top 25 .
Although there is a list of student comments in appendix B, the discussion of them in the text is very limited and there is no attempt an analysis. The Results sub-section (5.7) now contains a discussion of the student feedback and attempts to put them into a broader context.
Appendix C is largely advertising the programme, a couple of examples would suffice to illustrate the capabilities. Hand painting is very much an additional effort and not part of the model production process. Reduced to 4 images.
Appendix D is interesting and has some useful guidance, but is not well written and should be considered more effectively in the main body of the text. I think a ‘Best practice’ guide needs to be a bit more comprehensive, especially about issues such as suitable terrains, vertical scaling, size of model, etc. Some of Appendix D was integrated into a section on terrain printing guidelines (3.2), however some aspects are only relevant for experts in 3D printing (e.g. the choice of fill patterns or combing parameters) or address very specific scenarios (e.g. drawing/painting or gluing).These aspects were left in the appendix, which is now no longer using bullet text.
Reviewer 2 Report
Thank you for the paper.
Paper is very interesting and brings a useful application to the field. From the 3D printing community side, I see big potential in the paper.
Nevertheless from the scientific side, several parts of the papers are not necessary and the paper is very long. It's sometimes more like a manual to the software than a research paper. I will omit some information technologies for 3D printing as this is widely available. I will definitely more focus on some key aspect and better formulate the research questions. Which are hidden in the text.
The user cases are also an important part of research but not mentioned in the abstract. Concerning this, the rewriting of the abstract will be beneficiary to the readers as not all mentioned information in the paper is mentioned in the abstract. My first thoughts after reading abstract were that paper summarizes the all information acquired from the web from users.
From my perspective, the authors would like to present all activities connected with the Touch Terrain project in one paper. They also mention BVI users, geography etc.
My personal recommendation is to shorten the paper and focus more on some particular aspect of the research more in detail.
A literature review can be also deeper. There are exist more approach to generate a 3D model for 3D printing or other approaches on how to generate meshes like triangulation etc. Authors do not mention QGIS DEM3D or libraries like threejs.
Author Response
Response to reviewer 2 (in bold)
Paper is very interesting and brings a useful application to the field. From the 3D printing community side, I see big potential in the paper.
Nevertheless from the scientific side, several parts of the papers are not necessary and the paper is very long. It's sometimes more like a manual to the software than a research paper. I will omit some information technologies for 3D printing as this is widely available. We believe that the FDM 3D printing process, especially with regard to 3D printing terrain, is not yet sufficiently known to the typical reader of this journal. However, we have re-worked and shortened this section and integrated material from Appendix D (guidelines and heuristics) into a new sub-section which should be more relevant for readers already familiar with the basics of 3D printing.
I will definitely more focus on some key aspect and better formulate the research questions. Which are hidden in the text. The research questions are now spell out in the intro. “… to answer the following research questions: Who are our users? For what purpose are terrain models used for? What areas on the globe are most attractive for 3D terrain printing?”
The user cases are also an important part of research but not mentioned in the abstract. Concerning this, the rewriting of the abstract will be beneficiary to the readers as not all mentioned information in the paper is mentioned in the abstract. My first thoughts after reading abstract were that paper summarizes the all information acquired from the web from users.
The abstract does mention use cases (“Models are being created for many different use cases, including education, research, outdoor activities and crafting mementos”), however, I screwed up when entering the MDPI review form – I just noticed that its abstract field only show the last of the three abstract paragraphs!
From my perspective, the authors would like to present all activities connected with the Touch Terrain project in one paper. They also mention BVI users, geography etc.
My personal recommendation is to shorten the paper and focus more on some particular aspect of the research more in detail. Paper was shortened from 40 to 32 pages. Taking out much of the technological details, refocused the paper more towards the usage description and the ISU teaching experience.
A literature review can be also deeper. More than 10 additional references were added to the literature review.
There are exist more approach to generate a 3D model for 3D printing or other approaches on how to generate meshes like triangulation etc. Authors do not mention QGIS DEM3D or libraries like threejs. References to software such as QGIS DEM3D, threejs, Terrain2STL and The Terrainator (which are offer similar, but less feature rich functionality) were added to the intro. Pointed to extrusion as the primary approach for creating 3D mesh models from a DEM.
Reviewer 3 Report
Modern topic, well-structured paper, clear definition of the content and applications.
Keywords are correctly used for indexing the paper.
The article looks more like an advertisement than a scientific one, for example paragraph in 137-142 lines (page 3).
Point 2 “Background” is well known textbook literature.
The article is too long. The technical details as these ones on pages 11, 12, 13…. - could be reduced, this is more information for the user’s guide for Touch Terrain.
The cited literature is not enough. Appendix A does not support any scientific topic. The same is the problem with Table 1- how other researches, scientist, students will use this result?
FIG. 12., 13 - Equal area projection maps are better to be used for statistical data presentation. Fig 14: The map is designed without any cartographical principles of objects / phenomena representation.
Recommendations: The content could be reduced; and more cited papers could be added. The scientific content is enough for such kind of journal. Will be good if authors find new areas of applications of Touch Terrain.
Author Response
Response to reviewer 3 (in bold)
Modern topic, well-structured paper, clear definition of the content and applications.
Keywords are correctly used for indexing the paper.
The article looks more like an advertisement than a scientific one, for example paragraph in 137-142 lines (page 3). Paragraph was removed. Overly evangelizing language was toned down.
Point 2 “Background” is well known textbook literature.
The article is too long. Reduced from 40 pages to 32. The technical details as these ones on pages 11, 12, 13…. - could be reduced, this is more information for the user’s guide for Touch Terrain. This was pointed out by all reviewers. The amount of technical details was strongly reduced.
The cited literature is not enough. Added more than 10 additional references.
Appendix A does not support any scientific topic. The function of this table is merely to indicate which areas where most often searched for and which were therefore, by extension, interesting places to print, hence its relegation to the Appendix. This was clarified and the table was shortened to the top 25 places.
The same is the problem with Table 1- how other researches, scientist, students will use this result? This section was re-written and shortened for clarity and Table 1 was simplified. I now forms the basis for a rudimentary workflow analysis.
FIG. 12., 13 - Equal area projection maps are better to be used for statistical data presentation. Fig 14: The map is designed without any cartographical principles of objects / phenomena representation. These are limitations of the online mapping system, which also only support the Web Mercator projection. We stipulate that the interactive exploration of these online maps is more rewarding to the readers and that it will compensate for the failings of the static maps shown here (which are simply screenshots of the online maps).
Recommendations: The content could be reduced; and more cited papers could be added. Demonstrable efforts were made to satisfy these recommendations.
The scientific content is enough for such kind of journal. Will be good if authors find new areas of applications of Touch Terrain.
Round 2
Reviewer 1 Report
The authors are to be commended by taking on board the comments of the reviewers where practical, and improving the focus and presentation. It is now much more appropriate for a scientific journal and should be of interest to many readers.
A few specific comments:
Line 204 ‘In recent years …’ would be better
L 417/8 awkward grammar – ‘… z-scale is large enough when printed to result in …’
L605 Events can be used to analyze …
Fig 12 – the relabelling the x axis is a significant improvement - it now is much more readable
References – I have to say, I don’t like Cartographica being reduced to Cartogr. If anything this main element of the title should be retained in full and the qualifying part more condensed or removed.
Author Response
Line 204 ‘In recent years …’ would be better Done
L 417/8 awkward grammar – ‘… z-scale is large enough when printed to result in …’ Changed to: ... z-scale is large enough to result in a reasonably large number of layers. Printing at least 50 layers ensures that the model will show a good level of terrain details.
L605 Events can be used to analyze … Done
Fig 12 – the relabelling the x axis is a significant improvement - it now is much more readable
References – I have to say, I don’t like Cartographica being reduced to Cartogr. If anything this main element of the title should be retained in full and the qualifying part more condensed or removed. Well, that's not really our doing, we are using a reference formatting style that is officially sanctioned by MDPI. However, I have manually changed Cartogr. Int. J. Geogr. Inf. Geovisualization to Cartographica: Int. J. Geogr. Inf. Geovis. I hope that captures the spirit of your argument!
