Open research questions in 3D Shape Modeling and Analysis

What the literature leaves open

• High-resolution voxels are expensive to process. Point cloud processing remains computationally intensive. Existing point cloud processing pipelines incorporate a sampling step, which can lead to loss of information.

  A Systematic Survey on Deep Learning Architectures for Point Cloud Classification and Segmentation · 2026 · DOI

• Developing more sophisticated attention mechanisms specifically tailored to the unique characteristics of point clouds. Investigating sparse attention mechanisms, point-cloud-specific tokenizations, and quantized architectures to reduce memory or compute while maintaining performance. Exploring the use of multi-modal frameworks and state-space models for point cloud processing.

  A Systematic Survey on Deep Learning Architectures for Point Cloud Classification and Segmentation · 2026 · DOI

• Point cloud segmentation and editing require annotated training data or are limited to geometric primitives. LLM agents for spatial reasoning and editing are not fully validated. Repeated runs of LLM agents might yield different outcomes due to non-determinism.

  LLM-Supervised Point Cloud Processing: From Unsupervised 3D Scene-Graph Generation to Interactive Scene Manipulation · 2026 · DOI

• The method is limited by the amount of available working memory. The method is designed for in situ training, which can be challenging due to the sequential nature of the process.

  In situ training of implicit neural compressors for scientific simulations via sketch-based regularization · 2026 · DOI

• There is a need for performant compression methods that can operate in situ with the simulation. There is a lack of methods that can prevent catastrophic forgetting in in situ training.

  In situ training of implicit neural compressors for scientific simulations via sketch-based regularization · 2026 · DOI

• Despite the remarkable progress of modern manifold-aware encoders and generative transport models in geometric representation learning, a fundamental objective-geometry mismatch remains underexplored.

  RODR: Riemannian Orthogonally Decoupled Regularization for Disentangled Manifold Representation · 2026

• Such settings are insufficient for real-world applications with large scenes and severe inter-person occlusions.

  Multiview Multi-Person Human Mesh Recovery Under Large Scenes with Occlusions · 2026

• While density-robust circular coordinates have recently been developed, the extension to spherical coordinates remains an open challenge: unlike the circular case, spherical coordinates are obtained through a nonlinear variational problem for sphere-valued maps, to which existing density-correction mechanisms are not directly applicable.

  Density-Robust Spherical Coordinates from Persistent Cohomology · 2026

• These results support robust rigid initialization under controlled synthetic surface loss on FaceScape; generalization to real sensor-acquired point clouds remains to be established.

  Dual-Expert Landmark Localization in 3D Facial Point Clouds Under Controlled Synthetic Local Surface Loss for Rigid Initialization · 2026 · DOI

• We conclude with practical lessons and open challenges in data, benchmarks, and trustworthy integration.

  A survey of AI methods for geometry preparation and mesh generation in engineering simulation · 2026 · DOI

• The definition of spatial autocorrelation is not clear, - The literature is replete with spatial simulations where the stopping rule is not clear

  Simulating Two‐dimensional Autocorrelated Surfaces · 1983 · DOI

• The inherent sparsity and irregularity of point cloud data. The lack of effective interaction between point cloud features and geometric information. The need to handle sparse and non-uniform point clouds.

  Hgca-net: hybrid feature affine and geometric contextual aggregation networks for point cloud semantic segmentation · 2026 · DOI

• Existing methods fail to fully exploit the intrinsic geometric structure of point clouds. The lack of effective interaction between point cloud features and geometric information.

  Hgca-net: hybrid feature affine and geometric contextual aggregation networks for point cloud semantic segmentation · 2026 · DOI

• The present implementation assumes a single material property (constant stiffness) and uniform shell thickness. The approach does not explicitly take into account the distance between the air release point and each vertex. The method is limited to real-time applications.

  Physically plausible balloon dynamics via position-based constraints and geodesic-weighted forces · 2026 · DOI

• Incorporating temperature-dependent elasticity. Modeling multiple material properties and non-uniform shell thickness. Exploring applications in various fields such as advertising, exhibitions, and interactive content.

  Physically plausible balloon dynamics via position-based constraints and geodesic-weighted forces · 2026 · DOI

• The paper does not fully cover adverse-weather or cross-sensor deployment. The stress test is limited to two controlled stressors: sparsification and range truncation.

  GDPNet: a hybrid GNN-Transformer with position–density-modulated attention for 3D point cloud semantic segmentation · 2026 · DOI

• Investigating the application of GDPNet to other 3D point cloud tasks. Exploring the use of GDPNet in adverse-weather or cross-sensor deployment. Improving the efficiency of GDPNet for large-scale 3D point cloud data.

  GDPNet: a hybrid GNN-Transformer with position–density-modulated attention for 3D point cloud semantic segmentation · 2026 · DOI

• The challenge of complete-to-partial registration in image-guided liver surgery. The limitation of prior registration methods based on point cloud completion. The need for a more accurate and robust registration framework.

  PCReg: a coarse-to-fine registration framework using point cloud completion for intraoperative liver deformation correction · 2026 · DOI

• Reliability and quality assurance remain challenging tasks in Fused Filament Fabrication 3D printing. There is a need for a method to generate a high-resolution reconstruction of the printed object.

  Camera based in situ layer segmentation and object reconstruction for digital twins in FFF 3D printing · 2026 · DOI

• References Future work should focus on addressing the identified limitations. To reduce the added print time, several strategies can be pursued: implementing a non-blocking image acquisition workflow, optimizing camera movements, capturing only regions with newly deposited material, enlarging the camera’s field of view, or replacing the scanning camera with a single fixed high-resolution camera. For segmentation, improvements include generating more accurate ground truth Abadi, M. (2016). Tensorflow: learning functions at scale. In: Proceedings of the 21st ACM SIGPLAN International Conference on Functional Programming. ACM, New York, NY, USA. https:// doi.org/10.1145/2951913.2976746 AbouelNour, Y., & Gupta, N. (2023). Assisted defect detection by in-process monitoring of additive manufacturing using optical imaging and infrared thermography. Additive Manufacturing, 67, Article 103483. https://doi.org/10.1016/j.addma.2023.103483 Ahlers, D., Wasserfall, F., Hörber, J., & Zhang, J. (2023). Automatic in-situ error correction for 3d printed electronics. Additive Man- 123 ufacturing Letters, 7, Article 100164. https://doi.org/10.1016/j. addlet.2023.100164 Brion, D. A. J., & Pattinson, S. W. (2022). Generalisable 3d printing error detection and correction via multi-head neural networks. Nature Communications. https://doi.org/10.1038/s41467- 022-31985-y Bowoto, O. K., Zahedi, S. A., & Chong, S. (2023). Enhancing dimensional accuracy in 3d printing: A novel software algorithm for real-time quality assessment. The International Journal of Advanced Manufacturing Technology, 129(7–8), 3435–3446. https://doi.org/10.1007/s00170-023-12543-2 Caltanissetta, F., Dreifus, G., Hart, A. J., & Colosimo, B. M. (2022). In-situ monitoring of material extrusion processes via thermal videoimaging with application to big area additive manufacturing (baam). Additive Manufacturing, 58, Article 102995. https:// doi.org/10.1016/j.addma.2022.102995 Charalampous, P., Kostavelis, I., Kopsacheilis, C., & Tzovaras, D. (2021). Vision-based real-time monitoring of extrusion additive manufacturing processes for automatic manufacturing error detection. The International Journal of Advanced Manufacturing Technology, 115(11–12), 3859–3872. https://doi.org/10.1007/ s00170-021-07419-2 Delli, U., & Chang, S. (2018). Automated process monitoring in 3d printing using supervised machine learning. Procedia Manufacturing, 26, 865–870. https://doi.org/10.1016/j.promfg.2018.07.111 Fayazbakhsh, K., Movahedi, M., & Kalman, J. (2019). The impact of defects on tensile properties of 3d printed parts manufactured by fused filament fabrication. Materials Today Communications, 18, 140–148. https://doi.org/10.1016/j.mtcomm.2018.12.003 Girard, J., & Zhang, S. (2025). Fast error detection method for additive manufacturing process monitoring using structured light three dimensional imaging technique. Optics and Lasers in Engineering, 184, Article 108609. https://doi.org/10.1016/j.optlaseng. 2024.108609 Jeong, H., Kim, M., Park, B., & Lee, S. (2017). Vision-based realtime layer error quantification for additive manufacturing. In: Volume 2: Additive Manufacturing; Materials. MSEC2017. American Society of Mechanical Engineers, Los Angeles, California, USA. https://doi.org/10.1115/msec2017-2991 Jin, Z., Zhang, Z., & Gu, G. X. (2019). Autonomous in-situ correction of fused deposition modeling printers using computer vision and deep learning. Manufacturing Letters, 22, 11–15. https://doi.org/ 10.1016/j.mfglet.2019.09.005 Kuipers, T., Doubrovski, E. L., Wu, J., & Wang, C. C. L. (2020). A framework for adaptive width control of dense contour-parallel toolpaths in fused deposition modeling. Computer-Aided Design, 128, Article 102907. https://doi.org/10.1016/j.cad.2020.102907 Kim, C., Espalin, D., Cuaron, A., Perez, M. A., MacDonald, E., & Wicker, R. B. (2018). Unobtrusive in situ diagnostics of filamentfed material extrusion additive manufacturing. IEEE Transactions on Components, Packaging and Manufacturing Technology, 8(8), 1469–1476. https://doi.org/10.1109/tcpmt.2018.2847566 Liu, C., Law, A. C. C., Roberson, D., & Kong, Z. J. (2019). Image analysis-based closed loop quality control for additive manufacturing with fused filament fabrication. Journal of Manufacturing Systems, 51, 75–86. https://doi.org/10.1016/j.jmsy.2019.04.002 Li, L., McGuan, R., Isaac, R., Kavehpour, P., & Candler, R. (2021). Improving precision of material extrusion 3d printing by in-situ monitoring and predicting 3d geometric deviation using conditional adversarial networks. Additive Manufacturing, 38, Article 101695. https://doi.org/10.1016/j.addma.2020.101695 Leng, S., McGee, K., Morris, J., Alexander, A., Kuhlmann, J., Vrieze, T., McCollough, C.H., Matsumoto, J.: Anatomic modeling using 3d printing: quality assurance and optimization.

  Camera based in situ layer segmentation and object reconstruction for digital twins in FFF 3D printing · 2026 · DOI

• The generated thermal point cloud heavily relies on accurate RGB reconstruction and scale estimation. Thermal Infra-Red (TIR) images are inherently 2D and have limited spatial resolution, narrow Field of View (FoV), and poor geometric fidelity.

  Automatic generation of thermal point clouds from TIR images and LiDAR point clouds · 2026 · DOI

• Current methods suffer from macro-topological inconsistency and micro-geometric discontinuity. The potential of SPD manifolds as a regularization mechanism in text-to-3D generation tasks remains underexplored.

  MOC-3D: Manifold-Order Consistency for Text-to-3D Generation · 2026 · DOI

• In this paper, we propose MOC-3D, a novel synergistic optimiza- tion framework addressing the issues of macro-topological incon- sistency and micro-geometric discontinuity in existing text-to-3D generation methods. Built upon ScaleDreamer, this framework in- novatively introduces dual consistency constraints. At the macro level, the Semantic View-Order Constraint Module leverages CLIP priors to construct a monotonicity rank constraint. This injects ex- plicit global structural guidance into the optimization process, effec- tively rectifying multi-head topological anomalies such as the Janus problem. At the micro level, the Manifold-based Feature Continuity Module models multi-view features as points on the Symmetric Positive Definite (SPD) manifold. Utilizing Riemannian geometric metrics, it constrains the smooth evolution of feature distributions at a statistical level, effectively suppressing high-frequency arti- facts and geometric discontinuities induced by gradient noise. To comprehensively validate the effectiveness of our method, we de- signed a systematic experimental protocol and drew the following conclusions. First, on test prompts covering diverse topological and textural characteristics, MOC-3D outperforms mainstream base- lines (e.g., DreamFusion, ScaleDreamer, and Hunyuan3D) in key metrics such as Semantic Consistency (CLIP Score) and Perceptual Quality (LPIPS), effectively resolving multi-face effects and texture inconsistencies. Second, ablation studies confirm the effectiveness of the dual constraints, indicating that the Semantic View-Order Constraint effectively rectifies macro-topological errors, while the Manifold-based Feature Continuity Constraint is responsible for smoothing micro-texture details. Their synergy achieves an optimal balance between structure and texture. Third, in extended experi- ments on specific cultural style scenarios, MOC-3D demonstrates exceptional robustness. It successfully overcomes the limitations of baseline models in handling high-frequency repetitive textures and hybrid geometric structures, effectively avoiding texture over- smoothing or structural proportion misalignment caused by single- module constraints, thereby realizing high-fidelity generation of complex hybrid topologies. ICMR ’26, June 16–19, 2026, Amsterdam, Netherlands Fan, Cheng, Yang, et al.

  MOC-3D: Manifold-Order Consistency for Text-to-3D Generation · 2026 · DOI

• Further improvement of the proposed method to handle extremely low point density or very large density variations. Application of the proposed method to other 3D point cloud segmentation tasks. Investigation of the use of other types of handcrafted features to improve instance segmentation accuracy.

  SPPSFormer: High-Quality Superpoint-Based Transformer for Roof Plane Instance Segmentation from Point Clouds · 2026 · DOI

• Existing Superpoint Transformers suffer from limited performance due to low-quality superpoints. Traditional plane instance segmentation methods have limitations, such as relying on prior knowledge and manual parameter tuning.

  SPPSFormer: High-Quality Superpoint-Based Transformer for Roof Plane Instance Segmentation from Point Clouds · 2026 · DOI

Most-cited papers in 3D Shape Modeling and Analysis

• SuGaR: Surface-Aligned Gaussian Splatting for Efficient 3D Mesh Reconstruction and High-Quality Mesh Rendering · 2024 · 451 citations
• GaussianAvatars: Photorealistic Head Avatars with Rigged 3D Gaussians · 2024 · 251 citations
• Feature 3DGS: Supercharging 3D Gaussian Splatting to Enable Distilled Feature Fields · 2024 · 211 citations
• 3DGS-Avatar: Animatable Avatars via Deformable 3D Gaussian Splatting · 2024 · 179 citations
• CLAY: A Controllable Large-scale Generative Model for Creating High-quality 3D Assets · ACM Transactions on Graphics · 2024 · 171 citations
• GaussianAvatar: Towards Realistic Human Avatar Modeling from a Single Video via Animatable 3D Gaussians · 2024 · 169 citations
• Animatable Gaussians: Learning Pose-Dependent Gaussian Maps for High-Fidelity Human Avatar Modeling · 2024 · 153 citations
• SplattingAvatar: Realistic Real-Time Human Avatars With Mesh-Embedded Gaussian Splatting · 2024 · 147 citations
• HUGS: Human Gaussian Splats · 2024 · 138 citations
• OneFormer3D: One Transformer for Unified Point Cloud Segmentation · 2024 · 136 citations

Most recent work

• Improved Point Cloud Representation via a Learnable Sort–Mix–Attend Mechanism · Sensors · 2026
• A novel structured mesh generation method based on physics-guided Kolmogorov superposition network · Engineering Analysis with Boundary Elements · 2026
• Efficient construction of feature-controlled curves with geometric continuity · Computers & Graphics · 2026
• MyMesh: General purpose, implicit, and image-based meshing in Python · The Journal of Open Source Software · 2026
• A Diffusion Framework based on Prompt-Driven Masks for Part-level 3D Editing · Journal of Computational Design and Engineering · 2026
• Robust Real-Time Garment Fitting from 3D Point Clouds with Physics-Guided Uncertainty and Reliability Monitoring · Journal of Fiber Bioengineering and Informatics · 2026
• Automatic and robust wrapping for complex defective model · Engineering Computations · 2026
• Large-Scale Airborne LiDAR Point Cloud Building Extraction Based on Improved Voxelized Deep Learning Network · Buildings · 2026
• Object shape differentiation and texture rendering for neural implicit SLAM · Machine Vision and Applications · 2026
• Next Bit Prediction: A Unified Lossless and Lossy Point Cloud Geometry Compression Framework · IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026

Related topics in Engineering