Abstract Creation of collision objects for 3D models is a time‐consuming task, requiring modelers to manually place primitives such as bounding boxes, capsules, spheres, and other convex primitives to approximate complex meshes. While there has been work in automatic approximate convex decompositions of meshes using convex hulls, they are not practical for applications with tight performance budgets such as games due to slower collision detection and inability to manually modify the output while maintaining convexity as compared to manually placed primitives. Rather than convex decomposition with convex hulls, we devise an approach for bottom‐up decomposition of an input mesh into convex primitives specifically for rigid body simulation inspired by quadric mesh simplification. This approach fits primitives to complex, real‐world meshes that provide plausible simulation performance and are guaranteed to enclose the input surface. We test convex primitive decomposition on over 60 models from Sketchfab, showing the algorithm's effectiveness. On this dataset, convex primitive decomposition has lower oneway mean and median Hausdorff and Chamfer distance from the collider to the input compared to V‐HACD and CoACD, with less than one‐third of the complexity as measured by total bytes for each collider. On top of that, rigid‐body simulation performance measured by wall‐clock time is consistently improved across 24 tested models.

3D artists blend geometry and texture to craft objects and their appearance, using both geometric and image processing tools. Yet, it is difficult for geometry processing to account for textures in the standard UV with texture maps representation. When textures are represented as vertex colors though, we show texture-aware geometry processing becomes straightforward. Unfortunately, most textured meshes are represented with UV parameterizations, so in practice most geometry processing tools simply ignore texture. This scarcity of vertex colors meshes is remedied using a new remeshing approach that hoists textures to vertex colors through per-triangle remeshing, while preserving input appearance. This new remeshing takes meshes with UVs and texture maps and generates vertex color meshes, which are then used to show how vertex colors can be incorporated into geometry processing through texture-aware modifications to multiple applications. Specifically, we test texture-aware geometry processing on surface segmentation, Tutte parameterizations, surface texture processing algorithms such as edge-detection, and vector field operations with the texture gradient. For each application, texture information produces different and in some sense improved results as compared to geometry-only algorithms, or image processing algorithms performed in UV space.

Mesh reduction using quadric error metrics is the industry standard for producing level-of-detail (LOD) geometry for meshes. Although industry tools produce visually excellent LODs, mesh topology is often ruined during decimation. This is because tools focus on triangle simplification and preserving rendered appearance, whereas artists often produce quad dominant meshes with clean edge topology. Artist created manual LODs preserve both appearance and quad topology. Furthermore, most existing tools for quad decimation only accept pure quad meshes and cannot handle any triangles. The gap between quad and triangular mesh decimation is because they are built on fundamentally different operations, triangle simplification uses single edge collapses, whereas quad decimation requires that entire sets of edges be collapsed atomically. In this work, we demonstrate that single edge collapse can be used to preserve most input quads without degrading geometric quality. Single edge collapse quad preservation is made possible by introducing dihedral-angle weighted quadrics for every edge, allowing optimization to evenly space edges while preserving features. It is further enabled by explicitly ordering edge collapses with nearly equivalent quadric error in a way that preserves quad topology. In addition to quad preservation, we demonstrate that by introducing weights for quadrics on certain edges, our framework can be used to preserve symmetry and joint influences. To demonstrate our approach is suitable for skinned mesh decimation on triangle meshes, we show that QEM with attributes can preserve joint influences better than prior work. We implement and test our approach on 67 static and 19 animated meshes from Sketchfab. On both static and animated meshes, our approach consistently outperforms prior work with lower Chamfer and Hausdorff distance, while preserving more quad topology when present.

Maintaining memory-efficient 3D assets is critical for game development due to size constraints for applications, as well as runtime costs such as GPU data transfers. While most prior work on 3D modeling focuses on reducing triangle count, few works focus on reducing texture sizes. We propose an automatic approach to reduce the texture size for 3D models while maintaining the rendered appearance of the original input. The two core components of our approach are (1) overlapping identical UV charts and folding mirrored regions within charts through an optimal transport optimization, and (2) carving redundant and void texels in a UV-aware and texture-aware way without inverting the UV mesh. The first component creates additional void space, whereas the second removes void space, and their combination can greatly increase texels utilized by the UV mesh at lower texture resolutions. Our method is robust and general, and can process a 3D model with arbitrary UV layout and multiple textures without modifying the 3D mesh. We evaluate our approach on 110 models from the Google Scanned Object dataset and 64 models from Sketchfab. Compared to other approaches, ours has on average 1 to 3 dB PSNR higher rendering similarity and reduces pixelation in visual comparisons.

Level of detail has been widely used in interactive computer graphics. In current industrial 3D modeling pipelines, artists rely on commercial software to generate highly detailed models with UV maps and then bake textures for low-poly counterparts. In these pipelines, each step is performed separately, leading to unsatisfactory visual appearances for low polygon count models. Moreover, existing texture baking techniques assume the low-poly mesh has a small geometric difference from the high-poly, which is often not true in practice, especially with extremely low poly count models. To alleviate the visual discrepancy of the low-poly mesh, we propose to jointly optimize UV mappings during texture baking, allowing for low-poly models to faithfully replicate the appearance of the high-poly even with large geometric differences. We formulate the optimization within a differentiable rendering framework, allowing the automatic adjustment of texture regions to encode appearance information. To compensate for view parallax when two meshes have large geometric differences, we introduce a spherical harmonic parallax mapping, which uses spherical harmonic functions to modulate per-texel UV coordinates based on the view direction. We evaluate the effectiveness and robustness of our approach on a dataset composed of online downloaded models, with varying complexities and geometric discrepancies. Our method achieves superior quality over state-of-the-art techniques and commercial solutions.

Recent implicit neural rendering methods have demonstrated that it is possible to learn accurate view synthesis for complex scenes by predicting their volumetric density and color supervised solely by a set of RGB images. However, existing methods are restricted to learning efficient representations of static scenes that encode all scene objects into a single neural network, and they lack the ability to represent dynamic scenes and decompose scenes into individual objects. In this work, we present the first neural rendering method that represents multi-object dynamic scenes as scene graphs. We propose a learned scene graph representation, which encodes object transformations and radiance, allowing us to efficiently render novel arrangements and views of the scene. To this end, we learn implicitly encoded scenes, combined with a jointly learned latent representation to describe similar objects with a single implicit function. We assess the proposed method on synthetic and real automotive data, validating that our approach learns dynamic scenes – only by observing a video of this scene – and allows for rendering novel photo-realistic views of novel scene compositions with unseen sets of objects at unseen poses.