Vertex Chunk-Based Object Culling Method for Real-Time Rendering in Metaverse
Round 1
Reviewer 1 Report
There are abbreviations which have not been explained e.g. AABB
There are also some terms which are not explained e.g. “early Z termination”
The symbols of equation 1 are not explained directly after the provision of the equation making it difficult for the reader to follow the paper.
In section 3.2, the authors claim that they achieve data compression by storing the centre of the AABB and the radius to the point furthest away in the positive direction from the centre then the AABB can be restored. AABB typically requires two points to be stored e.g., (x_min,y_min,z_min) and (x_max, y_max, z_max) ie. 6 parameters in total. The authors propose to store the centre (x_c,y_c,z_c) and the distance r_max between the centre and the point (x_max, y_max, z_max) ie. 4 parameters. Since AABB is a rectangular parallelepiped it is not clear how you can restore it using only its centre and the maximum distance proposed by the authors. For example using the centre and the max distance you can restore a cube or a number of parallelepipeds by selecting one of tis dimensions (length, height, depth) to fit the distance from the centre and the other two to be smaller.
The experiments that the authors provide prove that their proposed algorithm is superior to state of the art algorithms. However, the authors have selected to test their algorithm in two specific videos with sphere shaped objects. Such a selection may provide an advantage to their algorithm since the data compression technique used may become inefficient in videos with objects of randomized shapes deviating from the sphere e.g., buildings, trees, cars, people, etc.
In this context, the authors cannot claim that their solution is better vs. the compared techniques but they can claim that their technique appears to perform better in videos with sphere shaped objects.
- The paper requires a careful editing review in order to handle various typo or syntax errors (e.g., “. metaverse”, “Figure 1.”, “per objects”, “this method can be calculated”
Author Response
Dear Reviewer,
Thank you for your valuable comments and insightful feedback on our manuscript. We greatly appreciate the time and effort you dedicated to reviewing our work. Your suggestions have significantly contributed to improving the quality and clarity of our research.
We have carefully considered each of your comments and have made the necessary revisions accordingly. The highlighted sections(yellow) in the attached Word document indicate the modifications made for better readability, while the blue text represents our responses to your comments.
We would like to express our gratitude for your meticulous review, which has undoubtedly enhanced the overall strength of our manuscript. Your expertise and attention to detail have been instrumental in shaping the final version of our research.
Once again, we sincerely thank you for your valuable input and for the opportunity to improve our work. We remain open to any further suggestions or recommendations you may have.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Authors,
I have reviewed your paper on the proposed method for object culling in large and complex environments such as the Metaverse. While the research design, methods, results, and conclusions are generally well-presented, I have identified a few areas that could be improved to strengthen the paper. Please consider the following comments and suggestions:
Although you have compared your method with a few existing techniques, it would be beneficial to ensure that these techniques represent an exhaustive list of available solutions in the market. Including a more comprehensive comparison with relevant techniques would provide a more convincing argument for the effectiveness of the proposed method.
The study could benefit from a broader presentation of the results on a variety of scenes. Demonstrating the performance of the proposed method across diverse scenarios would help to validate its applicability and robustness.
The literature context could be expanded to provide a stronger foundation for the research within the existing body of work on object culling and rendering techniques. This would allow the reader to better understand the significance of the proposed method and its contribution to the field.
It would be beneficial to include a more detailed description of Figure 6 in the text. Describing the key features and differences between the regular and irregular world rendering videos would help readers better interpret the results.
Addressing these points would help to improve the quality and impact of the paper. I recommend revising the manuscript to address these concerns, which would make it a stronger candidate for publication.
Author Response
Dear Reviewer,
Thank you for your valuable comments and insightful feedback on our manuscript. We greatly appreciate the time and effort you dedicated to reviewing our work. Your suggestions have significantly contributed to improving the quality and clarity of our research.
We have carefully considered each of your comments and have made the necessary revisions accordingly. The highlighted sections(yellow) in the attached Word document indicate the modifications made for better readability, while the blue text represents our responses to your comments.
We would like to express our gratitude for your meticulous review, which has undoubtedly enhanced the overall strength of our manuscript. Your expertise and attention to detail have been instrumental in shaping the final version of our research.
Once again, we sincerely thank you for your valuable input and for the opportunity to improve our work. We remain open to any further suggestions or recommendations you may have.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Dear Authors,
Thank you for submitting the revised version of your manuscript. I have carefully reviewed the changes made and I appreciate the efforts you have made to address the comments and suggestions provided. The manuscript has shown improvement, and I believe it has the potential to be accepted in its current form with some minor revisions.
The methods section needs further elaboration. Provide more detailed explanations of the data collection process, experimental setup, and analysis techniques. This will help readers better understand the research methodology.
However, I would like to emphasize the importance of implementing the suggested changes from both my previous review and the previous reviewer's feedback. Incorporating these recommendations would significantly enhance the overall quality and clarity of the manuscript. I encourage you to consider all the suggestions and make the necessary revisions carefully.
The results section should be presented in a more structured and organized manner. Consider using figures and graphs to present your findings more effectively. Also, provide clear explanations and interpretations of the results.
It is essential to acknowledge the limitations of your study. Discuss any potential constraints or challenges faced during the research process and how they may have impacted the results. This will help readers evaluate the validity and generalizability of your findings.
By addressing the remaining concerns and implementing the suggested improvements, I believe your manuscript will be more robust and will have a stronger impact on the readers. I appreciate your attention to detail and your commitment to improving the manuscript. Once again, thank you for the opportunity to review your work.
I look forward to seeing the final version of your manuscript.
Author Response
Response to Reviewer Comments
Thank you for taking your valuable time to review this paper.
To see the revised part in detail, please download the word response file.
Point 1: The methods section needs further elaboration. Provide more detailed explanations of the data collection process, experimental setup, and analysis techniques. This will help readers better understand the research methodology.
Response 1: To provide more clarity, we have described the original data composition prior to data compression in the method section.
Our method does not collect data. Based on the context, it is understood that you are asking to describe the input data for rendering, which will be used for rendering the given content.This section explains how the original data was structured before undergoing data compression.
Object culling is a necessary technique for rendering multiple objects in real-time on Metaverse platforms. However, since additional calculations are required to select objects to cull in the overall process, performance may be lower than with previous methods. To address this issue, the proposed method uses a Vertex-Chunk to reduce the total amount of input data required for culling operations, resulting in more efficient calculations in the graphics pipeline.
The input data consists of objects that make up the 3D world in the content. These data include three-dimensional coordinates and resources such as shaders or textures corresponding to materials. Since most objects are rendered individually, culling operations are essential to improve the overall performance of the rendering process.
Although we have already provided a detailed description of the experimental environment in the experimental results section based on the previous review, we understand that it may have been perceived as insufficient. Therefore, we have additionally described the methods used to demonstrate efficiency.
To demonstrate the effectiveness of the proposed method, experiments were conducted to compare it with existing methods. In order to verify real-time rendering of multiple objects in the metaverse, it is crucial to measure the performance of the client process responsible for rendering the screen. Therefore, in the proposed method, the performance of the proposed approach was measured in terms of frames per second (fps) to compare it with other acceleration methods.
Point 2: However, I would like to emphasize the importance of implementing the suggested changes from both my previous review and the previous reviewer's feedback. Incorporating these recommendations would significantly enhance the overall quality and clarity of the manuscript. I encourage you to consider all the suggestions and make the necessary revisions carefully.
Response 2: All the opinions of the reviewers are valuable to us. If there are any instances where we may have misunderstood the intentions of the reviewers and deviated from their intended direction, please let us know. As you mentioned, we incorporate all the points from the reviewers into the manuscript as they contribute to the qualitative improvement of the manuscript.
Point 3: The results section should be presented in a more structured and organized manner. Consider using figures and graphs to present your findings more effectively. Also, provide clear explanations and interpretations of the results.
Response 3: We have added a graph displaying real-time frame rates and accompanying explanations to the existing table.
Figure 7. (a) Video measuring FPS in a moving view of the regular world, (b) Video measuring FPS in a moving view of the irregular world. The horizontal axis represents the scene number, and the vertical axis represents the FPS displayed.
Figure 7 presents the results of FPS measurements in a moving view. The camera view used for the measurements allows vertical movement in a 3D space, rather than a walkthrough. Starting from the ground with scene 1, the camera gradually ascends vertically to encompass the entire world as the scene number increases. As the view moves farther from the ground, the number of objects that need to be displayed in each scene increases. Consequently, the number of culled objects decreases, resulting in an overall decline in FPS. The proposed method exhibits the best performance compared to existing methods in this scenario. This is attributed to the effective improvement of the existing methods' limitations, which involve performance degradation as the input data increases. The proposed method achieves this by utilizing vertex chunks in the pre-processing stage to efficiently reduce the number of objects required for computation, ensuring optimal efficiency.
Point 4: It is essential to acknowledge the limitations of your study. Discuss any potential constraints or challenges faced during the research process and how they may have impacted the results. This will help readers evaluate the validity and generalizability of your findings.
Response 4: It is considered highly important to discuss the limitations of the research. In the final section of the experiment, we have discussed the following limitations:
In Table 5, we compared the average performance of object placement in the near view, close to the ground where objects are placed, and the far view captured from a high altitude, such as an aerial view. In the near view, chunk-level culling is performed during preprocessing, resulting in a reduced number of polygons being input to the GPU compared to the conventional HROC, and occlusion causes multiple objects to be culled. Therefore, it shows relatively faster rendering speed under these observation conditions.
However, in the far view of this experiment, more objects come into the field of view at once compared to the near view, and occlusion occurs less frequently. As a result, there is a situation where most objects need to be rendered because there are fewer objects to be culled. For this reason, as shown in Table 5, it is possible to achieve a performance improvement of over 20% compared to HROC in the near view, but in the far view, it shows similar or slower speed compared to the basic methods of view frustum culling and visibility culling. This is because they perform visibility checks for the objects composing the scene, but most of them are not culled. However, unlike the latest techniques such as HROC or ROC, the method that pre-removes objects using view frustum culling and visibility culling shows a significant performance difference due to the small input data size. This demonstrates the efficient resolution of the issue of computational overhead in cases where occlusion occurs infrequently, which was a problem with conventional HROC.
Point 5: By addressing the remaining concerns and implementing the suggested improvements, I believe your manuscript will be more robust and will have a stronger impact on the readers. I appreciate your attention to detail and your commitment to improving the manuscript. Once again, thank you for the opportunity to review your work. I look forward to seeing the final version of your manuscript.
Response 5: I feel that the manuscript has improved significantly thanks to the positive reviews from the reviewers. I am grateful for conducting the reviews, and I will cherish the feedback and continue to pursue further research. Thank you very much.
Thank you very much for your kind review.
Author Response File: Author Response.pdf
