Procedural Mesh Manipulation for Virtual Pottery Simulation with Hand Tracking

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presents a practical system for real-time virtual pottery simulation, providing a valuable exploration of vertex-based versus B-spline-based mesh manipulation, alongside an effective procedural texturing pipeline. While the current work demonstrates clear application potential, addressing the following suggestions would further strengthen its technical rigor and overall presentation:

Abstract and Introduction Logic: It is recommended to adjust the narrative flow. The authors should consider introducing the specific limitations of existing interactive pottery systems and datasets before presenting the proposed B-spline and vertex-based solutions.

Quantitative Benchmarks: While Table 1 provides a good qualitative overview, it will be highly beneficial to introduce a quantitative benchmark table. The authors should consider detailing the average per-frame latency, the memory overhead of the BFS neighborhood map, and peak memory bandwidth consumption.

Edge Device Constraints: Considering the target application in museum VR environments, it is suggested to discuss how the system handles the memory bandwidth and thermal limitations in terms of edge devices (such as standalone VR headsets), especially given the high-frequency inputs from the Leap Motion sensor.

Figure Presentation and Layout: It is suggested to clarify the meaning of any unlabeled or blank elements within the Figure 1 (a) flowchart. Besides, ensure that all editing artifacts (such as visible cursors or selection highlights) are removed. Additionally, the authors should consider adjusting the overall layout to ensure that the dimensions and alignment of the various figures are consistent throughout the manuscript.

Comparison with State-of-the-Art: It is suggested to add a brief comparison in the Related Work section to evaluate the proposed pipeline against recent lightweight mesh deformation SOTA models to further highlight the system's advantages.

Comments on the Quality of English Language

 The English could be improved to more clearly express the research.

Author Response

Dear Sir or Madam,

Thank you very much for your time and the constructive feedback from your review.  It is greatly appreciated and it helps us improve the quality and usefulness of the publication. Please find bellow the answers to your comments and improvement suggestions.

- Abstract and Introduction Logic - we edited the introduction to adjust the narrative flow. We hope the purpose of the publication and the application of the research and development described in it are now clearer. We also updated and added new references to the previous work section.
- Quantitative Benchmarks - we replaced Table 1 with quantitative results from tests and evaluations performed on two hardware configurations.
- Edge Device Constraints - the application is actually not a VR application. It is intended to run (and runs in the museum) on a standard PC with a standard display (and an additional touch monitor for UI interaction). This is not detailed in the publication as its focus is on the two mesh manipulation / deformation methods.
- Figure Presentation and Layout - we edited the two diagrams and updated the layout of the subfigures. We didn't see a remaining cursor or selection highlight in the images. There's a circle around the bottom of the pottery vessel, not sure if that is what was meant in the feedback, but that one is actually part of the turn table texture.
- Comparison with State-of-the-Art - as mentioned above we updated the Previous work section. 

Also as the submission type is a brief report, according to the requirements we're allowed only two figures and one table (https://www.mdpi.com/about/article_types).

Attaching the compiled pdf of the new version.

Best wishes,

Dimo Chotrov

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a real-time virtual pottery simulation system based on hand tracking, procedural mesh deformation, vertex painting, and triplanar texture projection. The application context is interesting, especially in relation to interactive museum installations and educational use, and the comparison between a vertex-based deformation strategy and a B-spline-based approach is potentially relevant. However, in its current form, the manuscript remains closer to a well-developed application prototype than to a fully convincing scientific study. The main limitations concern the lack of quantitative evaluation, the limited methodological novelty, several technical ambiguities, the weak evidential value of the current figures and table, and some issues in the bibliography and overall presentation.

1. The central claim of the paper is that the vertex-based and B-spline-based approaches exhibit different trade-offs in realism, locality, smoothness, and computational efficiency. However, this comparison is not supported by sufficient quantitative evidence. Table 1 summarizes the differences only qualitatively, using statements such as “high computational cost”, “performance efficient”, or “easier to achieve a good looking regular shape”, but no actual runtime, frame rate, memory usage, smoothing cost, or scalability results are reported. Since performance and interaction characteristics are central to the contribution, the paper would be substantially stronger if the authors included objective numerical comparisons, for example average update time per frame, FPS, impact of mesh resolution, cost of Laplacian smoothing, and memory overhead of the neighborhood structure.
2. The two main ingredients of the system—local mesh deformation with Laplacian smoothing and spline-based global shape control—are established techniques. The contribution of the manuscript appears to lie primarily in their integration into a virtual pottery application with hand tracking and procedural texturing, rather than in a substantially new modeling or deformation method. This is not necessarily a problem, but then the manuscript should position itself more clearly as an application-oriented comparative study and support its conclusions with stronger evaluation. At present, the balance between technical contribution and application development remains unclear.
3. The conclusions drawn in Section 4 are plausible, but they are not demonstrated rigorously enough. For example, the claim that the vertex-based method is more realistic but more computationally expensive, while the B-spline method is more stable and efficient, is reasonable, yet no experimental evidence is provided beyond qualitative visual examples and the narrative summary in Table 1. Similarly, claims regarding ease of obtaining aesthetically pleasing shapes or user learning curve would ideally require at least a small user-oriented evaluation, even if exploratory.
4. The methodological description still contains several technical ambiguities that weaken the reproducibility and rigor of the paper. The geometric construction of the initial cylindrical mesh is not described with sufficient precision, and as written it is not entirely clear whether the vertical sampling actually spans the intended full object height. This may reflect a notational inconsistency, but in its current form the description is ambiguous. Similar concerns apply to the B-spline section, where the definition of the quadratic spline and its knot sequence is not stated with enough rigor for the reader to fully reconstruct the method. In addition, one of the key interaction choices in the B-spline-based approach appears arbitrary: after ray intersection, the deformation reference is selected as the “first indexed vertex” of the intersected triangle. This is difficult to justify as a robust geometric criterion and should either be better motivated or replaced by a more meaningful strategy. Finally, although the smoothing stage in the vertex-based method is clearly important for the visual plausibility of the results, its actual impact is never quantified. At present, smoothing is treated as a visually helpful step rather than as a properly characterized component of the method, which is not sufficient given how central it is to the comparison.
5. Figures help the reader understand the workflow and visualize the qualitative outcomes of the two deformation methods. However, the flowcharts are quite schematic, several labels are small or hard to read, and the figures mainly play a demonstrative role. In particular, Figure 1 shows the visual effect of smoothing, but no objective metric is provided to quantify the improvement. Figure 2 is informative as an overview of the B-spline pipeline, but again functions more as a system illustration than as evidence.
6. Table 1 is not really a results table in the scientific sense. It is a qualitative discussion summary rather than a table of measured outcomes. If the authors wish to retain it, it should be complemented by at least one quantitative table in the main text reporting objective performance indicators. Otherwise, the strongest claims of the manuscript remain unsupported.
7. The paper would benefit from supplementary material containing detailed implementation parameters, extended benchmarks, additional visual comparisons, failure cases, and possibly a demonstration video. This would be especially helpful for a real-time interactive system. However, the key quantitative comparison between the two methods should still appear in the main paper, not only in supplementary files.

Comments on the Quality of English Language

The text is understandable, but the English is not yet publication-ready. There are several awkward or ungrammatical formulations, some stylistic repetitions, and minor editorial issues. A careful language revision would improve clarity and overall presentation.

Author Response

Dear Sir or Madam,  

Thank you very much for your time and the detailed constructive feedback you provided in your review. It is greatly appreciated and it helps us improve the quality and usefulness of the publication. Please find bellow the answers to your comments and improvement suggestions.  

1. We separated the results section in qualitative and quantitative results subsections, replacing the data in Table 1 with the results from the performance evaluation.
2. We have updated the Introduction and other sections to clarify the aim and focus of the publication.
3. As mentioned above we've tried to improve the presentation of the results. We were not able to conduct a survey comparing the ease of use of both methods. It is a general observation both on our side during the implementation as well as from the museum's experts who purchased the application. Actually the reason for implementing the B-Spline method was because experts from the museum said it is hard to achieve a good looking shape with the mesh-editing method which they didn't find suitable for the target audience - mainly children. Also the application is not yet live, because the museum is doing construction work, building a separate center for kids where the application will be hosted.
4. We updated the description regarding the first index vertex and added evaluation of the smoothing performance cost to the Results section. I didn't quite notice the ambiguity about the initial mesh that is being created. The vertical sampling does span the full height of the object. We used to have additional figures showing the initial meshes for both methods, but had to remove them - please see the answer to 5.
5. We agree that figures are very useful and benefit the ease of understanding. The type of our publication being a brief report per the requirements (detailed in https://www.mdpi.com/about/article_types) means it is allowed only "one or two figures and / or a table". We initially had more figures but had to decrease their number in order to fit those requirements. Hence also the somewhat strange combination of an activity diagram and screenshots. But we wanted to include both a diagram describing the respective algorithm as well as some visuals for what the results look like for each method.  Yes, the screenshots play a demonstrative role and we believe the visual quality improvement by the smoothing in the vertex-based method or the  aesthetically better results achieved by the B-Spline method are best evaluated visually.
6. We have changed the purpose of Table 1 and it now shows quantitative results.
7. Some of the mentioned, i.e. specific implementation parameters, benchmarks have been added to the text of the publication. Adding video recordings demonstrating how both methods work would indeed be beneficial. We have asked if it would be possible to upload such supplementary materials at a later point once we've been able to prepare those as an update to the publication.

Attaching the compiled pdf of the new version.

Best wishes,

Dimo Chotrov

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have made efforts to address the previous concerns, significantly strengthening the manuscript's technical rigor and overall clarity. For the final version, it would still be beneficial to briefly discuss how the observed performance metrics might translate to the specific thermal and compute constraints of standalone VR headsets typically used in museum environments, and perhaps add a brief sentence contextualizing the pipeline against recent SOTAs. Overall, the revisions are satisfactory, and I recommend acceptance.

Comments on the Quality of English Language

 The English could be improved to more clearly express the research.

Author Response

Dear Sir or Madam,

Thank you for your review. Please find below the answers to your comments. The new revision of the manuscript is also attached.

Best wishes.

Comment 1: For the final version, it would still be beneficial to briefly discuss how the observed performance metrics might translate to the specific thermal and compute constraints of standalone VR headsets typically used in museum environments
Response 1: The developed application is not intended (and not implemented) to run on VR headsets and we have not tested how it performs using such hardware.

Comment 2: Perhaps add a brief sentence contextualizing the pipeline against recent SOTAs
Response 2: We have added one extensive review paper to the related work and tried to clarify the positioning of the manuscript. Unfortunately we didn't find information about performance to compare to, for example neither RealPot nor PotteryVR provide information on per frame latency of their methods.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The revised manuscript has been substantially improved. The authors have addressed the main concern raised in R1 by adding a quantitative comparison between the vertex-based and B-spline-based approaches, including FPS, latency, smoothing cost, initialization time, and method-specific memory requirements. This addition makes the comparison more convincing and better supports the main conclusion that the B-spline approach is more suitable for the intended museum/educational application, while the vertex-based method provides more local and realistic deformation at higher computational and interaction cost. The manuscript is now more clearly framed as an application-oriented comparative study, which is appropriate for the scope of the work. I do not think that a further major revision is necessary. However, a few issues should still be addressed before publication.

First, the quantitative evaluation should be accompanied by a clearer description of the measurement protocol: how FPS and latency were measured, over what duration, with how many repetitions, using which profiling tools, under what interaction conditions, and whether the reported values are averages, representative ranges, or single observations.

Second, claims regarding ease of use, learning curve, and aesthetic quality should be softened, since the authors acknowledge that no formal user study was conducted. These statements should be explicitly presented as qualitative observations from the developers and museum experts, not as experimentally demonstrated user-experience results.

Finally, the B-spline interaction strategy based on selecting the first indexed vertex of the hit triangle remains somewhat arbitrary; if retained, it should be clearly described as an efficiency-driven implementation choice and acknowledged as a possible limitation.

Comments on the Quality of English Language

The manuscript still requires careful English editing.

Author Response

Dear Sir or Madam,

Thank you for your review and feedback. Please find below the answers to your comments. A revised version of the manuscript is also attached.

Best wishes

 

Comment 1: First, the quantitative evaluation should be accompanied by a clearer description of the measurement protocol: how FPS and latency were measured, over what duration, with how many repetitions, using which profiling tools, under what interaction conditions, and whether the reported values are averages, representative ranges, or single observations.
Response 1: Following your suggestion we have added respective clarification to the results description.

Comment 2: Second, claims regarding ease of use, learning curve, and aesthetic quality should be softened, since the authors acknowledge that no formal user study was conducted. These statements should be explicitly presented as qualitative observations from the developers and museum experts, not as experimentally demonstrated user-experience results.

Response 2: Following your suggestion we have softened the language and strictly clarified there has been no conducted user study and that the preference for the B-spline method is made based only on internal testing of the development team and museum experts.

Comment 3: Finally, the B-spline interaction strategy based on selecting the first indexed vertex of the hit triangle remains somewhat arbitrary; if retained, it should be clearly described as an efficiency-driven implementation choice and acknowledged as a possible limitation.

Response 3: We have edited the description (and the algorithm), removing the text related to first indexed vertex of the hit triangle - the spline control point is chosen directly depending on distance to the hit point (without determining hit triangle and a vertex from it). 

Author Response File: Author Response.pdf