You are currently viewing a new version of our website. To view the old version click .
MDPIMDPI
Search all articles
by
  • Naai-Jung Shih

Reviewer 1: Alfredo Restuccia Garofalo Reviewer 2: Anonymous Reviewer 3: Jiří Maxa

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript proposes an innovative approach to interpreting surrealism via artificial intelligence, encompassing the generation, reconstruction, and evaluation of surreal images and 3D models derived from conflicting prompts.”
From an engineering, computational, and 3D modelling standpoint, this review aims to make the work more transparent, replicable, and scientifically robust, by clarifying the generation methods, the quantitative evaluation set-up, and the consistency between results and claims. Comments focus on: (i) completeness of the AI-based workflow, (ii) reproducibility of measurements (area, volume, deviations), (iii) statistical analysis of the reported outcomes, and (iv) the balance between the artistic narrative and the technical–metric definition of “surreal identity”.

 

  • “…to assess the 3D reconstructed models as an exemplification of context…” — (at line 10)
    Please specify which 3D reconstruction algorithms were used (e.g., NeRF, 3D Gaussian Splatting, or a custom pipeline) and with which quantitative parameters (number of views/images, image resolution, camera/pose estimation accuracy, alignment error). As written, the sentence is conceptual but lacks technical detail on the geometric reconstruction method.
  • “…AI application flowchart in surrealism interpretation.” — (at line 117)
    Figure 2 is useful, yet the operational links between Dream Studio®, Stable Diffusion®, and RODIN® should be made explicit together with their data outputs (2D image, 3D mesh). Please indicate intermediate formats (e.g., OBJ/GLB/PLY), export/import steps, and the temporal sequence of the chained tools, so that the entire workflow can be replicated.
  • “…assessed in terms of structural details, from modified values of area and volume.” — (at line 190)
    The approach is promising, but it requires a normalisation model. Area/volume are compared after smoothing/decimation, yet the initial polygon/vertex count and the numerical tolerance are not reported. Please add a reference table (initial polygons; Δ% in area and volume per operation) and briefly discuss cumulative measurement error introduced by mesh operations.
  • “…assessment of 3D reconstructed surreal models.” — (at line 285)
    Figures 6–10 show area/volume comparisons. Please include a regression/correlation analysis (R²) to demonstrate the significance of the variations (e.g., area vs volume after Reduce Noise@5). Also clarify whether reported percentages come from single runs or averages over multiple trials, as this is essential to judge statistical robustness.
  • “…Surrealism should be revised and recursively reconstructable…” — (at line 483)
    The “3D AI Propagation” section is conceptually strong but technically brief. Please detail the export pipeline towards AR/VR, indicating the software (e.g., Sketchfab®, Unity, Blender) and the mapping between the 3D model and video/animation assets. A concise technical flow diagram would clarify the propagation from prompt to augmented scene.

        10.1016/j.autcon.2025.105986

10.1109/ICSECE61636.2024.10729491

Author Response

Dear Reviewer 1:

 

Thank you so much for this significant reviewing effort.

Your assistance is highly appreciated.

 

Best regards,

Naai-Jung Shih

 

 

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors
  1. What are the potential practical applications of the surreal images and corresponding 3D models exemplified in this study? Please elaborate on how these examples relate to the broader objectives of the work.
  2. Have the authors systematically evaluated the similarities and differences in how various AI models respond to the same prompts or prompt syntax when generating surreal images? Are the differences in image outputs presented in the manuscript random variations, or do they stem from the fundamental logic and architecture of the respective AI systems?
  3. For the “Relative values” defined on page 8 for image result comparisons, the authors should clearly describe and include explicit equations showing how these percentage values are calculated. The relationships between these values and other controlling parameters should also be quantitatively explained.
  4. In the conclusion section, the authors state, “There would be chained applications for a series of operations afterward. I found the novel approach of prompt context can create results to meet my needs.” These statements are vague and ambiguous. Please consider elaborating with concrete examples to clarify the intended meaning and implications.
  5. Throughout the manuscript, “The Goldfinch” is referenced multiple times for terminology definition and conceptual illustration. Given its recurring importance, it may be helpful for the authors to obtain permission to include detailed citations or images from The Goldfinch to help readers better understand the intended message and context.
  6. The list of keywords includes “augmented reality (AR),” yet its relevance to the authors’ work is not sufficiently discussed. Could the authors elaborate on how their work on AI-generated surreal images and 3D models relates to augmented reality in more detail? If AR is not a key focus of this study, the authors may consider removing this keyword for clarity and precision.

Author Response

Dear Reviewer 2:

 

Thank you so much for this significant reviewing effort.

Your assistance is highly appreciated.

 

Best regards,

Naai-Jung Shih

 

 

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

The authors produced a 26-page article, including references, on the topic of: Surreal AI: the generation, reconstruction, and assessment of surreal images and 3D models.

This study aims to interpret surrealism in artificial intelligence (AI) by evaluating its performance through 3D reconstruction. The authors carry out the interpretation using AI and apply conflicting stimuli to generate surrealistic images, as well as to reconstruct the 3D structural and visual details of the images.

I would recommend elaborating more in Chapters 2 and 3 on the analyses performed, providing clear justifications and, above all, a precise interpretation of the results. Conversely, the use of sentences posing rhetorical questions should be removed.

I would recommend revising the Conclusion to include a concise summary of the tangible results achieved. Theoretical analyses, if deemed important by the authors, should be relocated to the main body of the text.

The font in Figs. 2 and 4 needs to be more legible.
The descriptions of the graphs in Figs. 9 and 10 are difficult to read.

Comments on the Quality of English Language

English needs to be improved.

Author Response

Dear Reviewer 3:

 

Thank you so much for this significant reviewing effort.

Your assistance is highly appreciated.

 

Best regards,

Naai-Jung Shih

 

 

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

No more comments.

Author Response

Dear reviewer:

 

Thank you very much for the significant reviewing efforts. 

Your assistance is highly appreciated.

 

Best regards,

Naai-Jung Shih