iterative visual refinement

The absence of formal verification methods for G-code linear motion in 3D printing hinders assurance of geometric-code consistency. Method: This paper proposes a dimension-elevated semantic representation framework that parses G-code into sets of axis-aligned bounding boxes and their approximated point clouds, integrating geometric modeling with program analysis to enable invariant checking and differential testing across the manufacturing pipeline. Contribution/Results: The framework supports, for the first time, quantitative cross-slicer comparison (Cura vs. PrusaSlicer), error localization, and root-cause analysis of defects introduced during mesh repair (e.g., in MeshLab or Meshmixer). Evaluated on 58 real-world models, it efficiently detects slicing anomalies induced by small geometric features, exposes behavioral discrepancies among mainstream slicers, and identifies new errors inadvertently introduced during repair—thereby significantly enhancing the verifiability and reliability of end-to-end additive manufacturing pipelines.

为解决从正交视图生成CAD代码时出现的几何不一致问题，提出IterCAD框架，通过迭代修正方法逐步改进代码准确性。

This work addresses the lack of precise geometric validation in existing text-to-CAD generation methods, which often fail to correct dimensional inaccuracies. We propose a multi-agent collaborative framework that integrates inner-loop iterative correction of code execution errors with an outer-loop refinement combining exact geometric metrics from the OpenCASCADE kernel and global shape assessment via a vision-language model. Our approach uniquely unifies procedural geometric verification with visual-semantic judgment, yielding a retrieval-augmented generation (RAG) system that requires no fine-tuning and naturally evolves with CAD libraries. Evaluated on a newly curated benchmark of 100 multi-difficulty examples, our method achieves a 100% execution success rate, improves median IoU from 0.8085 to 0.9629, and reduces average Chamfer Distance from 28.37 to 0.74.

To address the low efficiency of manual CAD design review and the challenge of ensuring consistency between 3D models and reference images, this paper proposes ReCAD—the first end-to-end framework for automated CAD program review and repair. Methodologically, we construct CADReview, a large-scale paired dataset of CAD programs and corresponding images (20K+ samples), and design a customized architecture integrating multimodal understanding, program semantic parsing, and geometric spatial reasoning. We further introduce structured prompting and stepwise error correction to enable fine-grained error localization and program-level repair. Evaluated on the CADReview benchmark, ReCAD achieves a 32.7% improvement in error detection accuracy and a 28.4% increase in correction success rate over state-of-the-art multimodal large language models (MLLMs), demonstrating its effectiveness and practical potential for industrial design automation.

Large language models (LLMs) often generate CAD scripts that deviate from design intent due to geometric distortions or compilation failures. To address this, we propose CADCodeVerify—the first iterative vision-language verification framework tailored for CAD code generation. It leverages multimodal large models (e.g., GPT-4V) to automatically formulate verification-oriented visual questions, integrates CAD rendering, point-cloud Chamfer Distance (CD) evaluation, and feedback-driven multi-turn prompt optimization, enabling end-to-end refinement from natural language to high-fidelity, executable CAD code. Our contributions are threefold: (1) the first vision-closed-loop verification paradigm for CAD code generation; (2) CADPrompt—the first benchmark comprising 200 image-text–code triplets; and (3) on GPT-4, a 7.30% reduction in point-cloud CD distance and a 5.0% improvement in compilation success rate, significantly enhancing structural integrity and dimensional compliance.

This work addresses the degradation in generation quality observed in large language model (LLM)-driven CAD agents during iterative repair, primarily caused by the loss of user requirements, operation history, and failure evidence. To mitigate this, the authors propose a persistent tracking mechanism that jointly models user intent, modeling steps, failure evidence, and candidate outputs. The approach enables reliable repair through diagnosing faulty operations, performing bounded edit searches within their local dependency regions, and integrating execution validation with retention checks. For the first time, the method introduces persistent state tracking, localized dependency-aware editing, and reusable skill memory, significantly improving both repair success rates and geometric fidelity. Evaluated on DeepCAD, the system achieves state-of-the-art performance: ablating persistent state reduces repair success by nearly 50%, while removing local search doubles geometric error and API call count; pre-populating the skill library effectively lowers retry frequency, token consumption, and latency.

This study addresses the challenges of recovering sparse control cages from dense meshes, including inferring control requirements, missing feature curves, and difficulty in correcting initial results. To overcome these issues, this work proposes a modeler-inspired agent workflow that iteratively optimizes through planning and diagnosis phases. A stateful feedback mechanism is introduced to support repair or rollback decisions by incorporating semantic judgments into the pipeline. To ensure inspectability, the planner avoids directly generating vertices, instead integrating multi-view geometric evidence analysis with tool execution verification. Evaluated on a fixed benchmark queue, the proposed method outperforms automatic remeshing approaches across five metrics, reducing Chamfer-L1 to 0.443% and improving the F-score to 92.78%, thereby achieving highly controllable mesh-to-subdivision surface reconstruction.

This study addresses the limitations of recovering executable CAD programs from 3D meshes, specifically the difficulty in handling diverse modeling operations and correcting geometric errors. To this end, we propose a generative optimization framework that introduces a pioneering step-wise generation mechanism. Our method leverages large language models to predict modeling actions, integrating intermediate geometric states with an IoU-guided tree search to enable local editing and iterative refinement. Furthermore, we release ARCADE-1.5M, a large-scale dataset encompassing multi-operation CAD sequences. Extensive experiments demonstrate that our approach achieves state-of-the-art reconstruction accuracy across multiple benchmarks, yielding a relative IoU improvement of up to 87.2% on complex shapes while maintaining high program validity.

本文探讨了CAD中透明度渲染的问题，通过比较不同方法的性能与质量平衡，并提出针对特定应用场景的技术和未来研究方向。

Existing CAD generation methods struggle to simultaneously preserve modeling history, topological reference stability, and feature-level editability in cross-platform scenarios. This work proposes CADIR—an agent-oriented, executable intermediate representation that explicitly constructs a procedural graph encompassing operation sequences, parameter dependencies, constraints, and topological selections based on the OpenCASCADE (OCCT) geometric kernel. To enable faithful cross-platform model reconstruction, CADIR introduces a geometric signature matching mechanism. It is the first approach to support explicit procedural graph representations that allow editing across heterogeneous CAD backends. By integrating text- or image-driven procedural graph retrieval, CADIR demonstrates high-fidelity, editable reuse of complete models and substructures across FreeCAD, SolidWorks, and Fusion 360, enabling seamless subsequent modifications.