template-based projection

Existing methods for surface mesh reconstruction from multi-view images often rely on intermediate representations and post-processing steps, which can introduce geometric artifacts and fragmented structures. Direct optimization of explicit meshes, while appealing, faces significant challenges in adaptive topology handling and maintaining UV coordinate consistency. This work proposes ExMesh, a novel framework that, for the first time, seamlessly integrates discrete topological operations—such as vertex splitting and merging—into a continuous, differentiable optimization pipeline, enabling end-to-end explicit mesh reconstruction. By dynamically preserving UV coordinates during topology updates and supporting coarse-to-fine geometric refinement, ExMesh achieves a favorable balance among reconstruction accuracy, mesh compactness, and computational efficiency, substantially outperforming current state-of-the-art approaches.

This study addresses the challenge of simultaneously achieving topological consistency and high-fidelity geometric detail in dynamic mesh reconstruction by proposing a dual-mesh representation architecture. The method enables precise parameterization through barycentric coordinate mapping and surface-aligned 2D Gaussian splatting. Furthermore, it introduces a novel error-driven adaptive subdivision and merging mechanism that supports dynamic resolution adjustment while strictly preserving topological consistency. Experimental results demonstrate that the proposed approach achieves state-of-the-art geometric reconstruction accuracy while maintaining highly competitive photorealistic rendering quality.

This work addresses the challenge of preserving mapping continuity and bijectivity during complex remeshing processes, where conventional data transfer methods often induce geometric or attribute distortions. The authors propose a composite mapping framework based on local bijective atlases, enhanced by a Shared Scaffold structure that guarantees global bijectivity. The approach is generalized to support a variety of remeshing operations and, for the first time, enables the construction of bijective mappings on 3D tetrahedral remeshings by innovatively integrating Steinitz’s theorem with Maxwell–Cremona lifting theory. This framework facilitates precise tracking of geometric entities—including points, curves, and surfaces—across remeshing sequences, significantly improving fidelity in high-precision applications such as texture transfer and volumetric simulation.

Matching non-rigid 3D shapes with topological inconsistencies—such as holes, discontinuities, and noise—remains challenging, as existing methods rely on near-isometric or ARAP deformation assumptions that fail under common topological artifacts in real-world multi-view reconstructions. Method: We propose a topology-adaptive bidirectional matching framework that jointly optimizes a template mesh adapted to the target topology and its non-rigid alignments with both source and target shapes, eliminating reliance on global isometric or ARAP priors. Our approach synergistically optimizes ARAP regularization and bijectivity constraints, enabling high-fidelity, topology-robust correspondence estimation without data-driven priors. Results: Experiments demonstrate significant improvements over state-of-the-art learning-based and traditional methods under strong non-isometry and severe topological distortions, validating robustness and accuracy for real multi-view reconstruction matching tasks.

This work proposes a novel surface representation framework that addresses the lack of effective smooth interpolation methods for non-quadrilateral faces in arbitrarily topologized closed meshes. By integrating local polygonal quadratic interpolation with rational curve parameterization, the method constructs smoothly connected quadrilateral patches and introduces a specialized rational-curve-based parameterization strategy for triangular and general polygonal faces. Through sub-patch blending and surface stitching techniques, the approach achieves globally C¹-continuous, high-quality surface reconstruction. Notably, this is the first unified framework capable of effectively handling smooth interpolation across faces with arbitrary numbers of edges, significantly enhancing both the quality and flexibility of surface generation for complex-topology meshes.

Recovering editable, parameterized CAD construction sequences from geometric inputs such as meshes remains a fundamental challenge in design and manufacturing. This work proposes an IoU-driven hybrid optimization framework that, for the first time, formulates the reconstruction problem as structured CAD program optimization. By leveraging geometric feedback, the method iteratively fits and validates a rich set of parametric operations—including fillets and chamfers—within the procedural representation. The approach enables end-to-end image-to-CAD reconstruction across multiple modalities and significantly outperforms existing methods on established benchmarks, achieving superior performance in both volumetric IoU and Chamfer distance metrics. Moreover, it substantially reduces redundancy in the reconstructed programs, enabling efficient and high-fidelity recovery of complex CAD models.

This work addresses the challenge of efficiently integrating dynamic 3D reconstruction with physics simulation, which is hindered by the complexity of collision detection under changing mesh topologies. The authors propose a dual-representation framework that employs a fixed-topology mesh to enable efficient physical simulation while leveraging Gaussian splatting for high-quality rendering. To handle topological changes, they introduce strategies including vertex buffer updates, temporal correspondence tracking, and stencil projection. Their systematic evaluation—the first of its kind—demonstrates a 4.65× speedup in simulation compared to variable-topology baselines, albeit at the cost of a 65–80% reduction in geometric fidelity during topology transitions. These findings reveal a fundamental trade-off between high-fidelity reconstruction and physics-compatible mesh topologies.

Existing autoregressive methods for 3D mesh generation suffer from slow inference and error accumulation, making them ill-suited for real-time creation of high-quality meshes with artist-friendly topology. This work proposes the first native mesh generation framework based on flow matching: it employs a vertex-set mesh VAE to encode meshes into continuous latent representations and introduces a two-stage cascaded flow matching model that enables single-pass, parallel decoding from image to complete mesh. The approach eliminates the need for vertex quantization and welding, supports interactive generation (median time of 6 seconds), explicit control over face count, and multi-part asset modeling. It significantly outperforms current autoregressive baselines in both geometric fidelity and generation efficiency, achieving speedups of over an order of magnitude.

This work proposes a novel method for directly generating compact 3D meshes with artist-friendly triangle topology from geometric conditions such as signed distance fields. The key innovation lies in the introduction of a Nearest Vertex Field (NVF)—an implicit representation of surface mesh topology—combined with a latent flow matching model to generate this field. Structured, high-fidelity meshes are then produced end-to-end through NVF-driven region clustering followed by a topology-aware constrained Quadric Error Metric (QEM) simplification algorithm. Compared to existing learning-based approaches, the proposed method achieves substantially improved topological quality and generalization, reducing Chamfer Distance by 90% and accelerating inference by 8×.

This work addresses the high computational overhead incurred when dynamically edited tetrahedral meshes—subject to topological changes such as fracture, refinement, or merging—require full recomputation of solver states. The authors propose an exact streaming assembly method that leverages a pre-allocated superset mesh and a known sequence of edits to enable persistent incremental updates, thereby replacing global reassembly. This approach yields results mathematically equivalent to full reconstruction while preserving the solver, preconditioner, and time-stepping scheme unchanged. By maintaining structural continuity in unaffected regions, the method drastically reduces redundant computation. Experiments on 3D scenes with up to 460k elements demonstrate matrix update costs reduced by several orders of magnitude, end-to-end speedups of 1.37–1.61×, and up to a 76% reduction in per-frame simulation time for fracture scenarios—all without any deviation from baseline accuracy.