ActionMesh：基于时序 3D 扩散的动画网格生成
ActionMesh: Animated 3D Mesh Generation with Temporal 3D Diffusion
把预训练 3D 扩散模型沿时间轴扩展为「时序 3D 扩散」，并用时序 3D 自编码器把独立形状序列转成参考网格的形变，从而前馈式输出无骨骼、拓扑一致的动画网格；做 4D 生成或视频驱动建模的工作可直接复用这一「先出形状序列、再回归形变」的两段式框架。
原文依据 modify existing 3D diffusion models to include a temporal axis, resulting in a framework we dubbed "temporal 3D diffusion"
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论生成动画 3D 物体是许多应用的核心，然而多数先进工作在实践中的应用都较为困难，原因在于其设置受限、运行时间长或质量有限。我们提出 ActionMesh，一个以「动作中」的形式前馈式预测可直接用于生产的 3D 网格的生成模型。受早期视频模型启发，我们的核心洞见是改造现有 3D 扩散模型以纳入时间轴，由此得到一个我们称为「时序 3D 扩散」的框架。具体而言，我们首先改造 3D 扩散阶段，使其生成一串同步的潜变量，用以表示随时间变化且彼此独立的 3D 形状。其次，我们设计了一个时序 3D 自编码器，将一串独立形状转换为对某一预定义参考形状的相应形变，从而使我们能够构建动画。结合这两个组件，ActionMesh 可从单目视频、文本描述，甚至一个 3D 网格外加描述其动画的提示词等不同输入生成动画 3D 网格。此外，与先前方法相比，我们的方法速度快，且结果无需骨骼绑定、拓扑一致，因而能够支持快速迭代以及纹理贴图与重定向等无缝应用。我们在标准的视频到 4D 基准（Consistent4D、Objaverse）上评估了模型，在几何精度与时序一致性两方面均报告了最优性能，表明我们的模型能以空前的速度与质量交付动画 3D 网格。
英文原文摘要 1569 字符
Generating animated 3D objects is at the heart of many applications, yet most advanced works are typically difficult to apply in practice because of their limited setup, their long runtime, or their limited quality. We introduce ActionMesh, a generative model that predicts production-ready 3D meshes "in action" in a feed-forward manner. Drawing inspiration from early video models, our key insight is to modify existing 3D diffusion models to include a temporal axis, resulting in a framework we dubbed "temporal 3D diffusion". Specifically, we first adapt the 3D diffusion stage to generate a sequence of synchronized latents representing time-varying and independent 3D shapes. Second, we design a temporal 3D autoencoder that translates a sequence of independent shapes into the corresponding deformations of a pre-defined reference shape, allowing us to build an animation. Combining these two components, ActionMesh generates animated 3D meshes from different inputs like a monocular video, a text description, or even a 3D mesh with a text prompt describing its animation. Besides, compared to previous approaches, our method is fast and produces results that are rig-free and topology consistent, hence enabling rapid iteration and seamless applications like texturing and retargeting. We evaluate our model on standard video-to-4D benchmarks (Consistent4D, Objaverse) and report state-of-the-art performances on both geometric accuracy and temporal consistency, demonstrating that our model can deliver animated 3D meshes with unprecedented speed and quality.