用单次前向传播将视频扩散模型潜空间转为精确三角面片，实现可实时渲染的3D场景生成。
FLAT: Feedforward Latent Triangle Splatting for Geometrically Accurate Scene Generation
- 提出射线中心旋转参数化与新型窗口函数，解决三角面片生成中的梯度难题。
- 在标准评测中几何精度显著优于现有前向基线，视觉质量保持领先。
- 生成结果可轻量优化为游戏引擎可用的完整表面模型，适合实时应用。
从单张图像生成可探索的3D场景需要强大的生成先验和适用于下游任务的精确几何表示。当前视频扩散模型虽能生成高质量图像，并在潜空间隐式编码多视角结构，但现有的前向潜空间解码器通常输出体素化3D高斯，缺乏明确定义的表面，限制了其在仿真或标准图形管线中的使用。这促使我们探索可对齐表面的显式几何原语，不仅可渲染，且更接近真实几何资产。我们探究是否能将压缩的视频扩散潜变量直接映射为单次前向传播的显式表面原语。为此，我们提出FLAT，并首次证明可直接从视频扩散潜变量解码三角面片。相比解码3D高斯，预测平面原语更具挑战性，因对原语方向高度敏感，常导致梯度传播不佳。FLAT通过两项关键设计解决：基于射线中心的三角形旋转参数化，以及一种新型乘积窗口函数，改善可微分三角面片渲染中的梯度流动。在标准基准上，FLAT在几何精度上显著优于最先进的前向基线，同时保持竞争力的视觉质量。我们进一步展示，一个轻量级测试时优化步骤可将预测的三角面片集合转化为完全不透明、游戏引擎就绪的表示，支持实时渲染。通过在相同训练设置下对比3DGS、2DGS与三角面片解码变体，我们首次系统分析了前向场景生成中的表示权衡。项目页面见 https://flat-splat.github.io
原文摘要 · Abstract (English)
Generating explorable 3D scenes from a single image requires strong generative priors and accurate geometric representations suitable for downstream use. Current video diffusion models offer high-quality generation and implicitly encode multi-view geometric structure in latent space. However, existing feedforward latent scene decoders typically output volumetric 3D Gaussians that lack a well-defined surface, limiting their use in simulation or standard graphics pipelines. This motivates decoding surface-aligned primitives that are not only renderable but also closer to explicit geometric assets. We ask whether compressed video diffusion latents can be mapped directly to explicit surface primitives in a single pass. To this end, we introduce FLAT and, for the first time, show that triangle splats can be decoded directly from video diffusion latents. Compared with decoding 3D Gaussians, predicting flat primitives is notoriously more challenging due to high sensitivity to primitive orientations, oftentimes leading to poor gradient flow. FLAT solves with two key ingredients: a ray-centered rotation parameterization for triangle regression and a novel product window function that improves gradient flow during differentiable triangle rendering. On standard benchmarks, FLAT achieves significantly better geometric accuracy while maintaining competitive visual quality compared to state-of-the-art feedforward baselines. We further show that a lightweight test-time refinement step converts the predicted triangle soup into a fully opaque, game-engine-ready representation that supports real-time rendering. By evaluating 3DGS, 2DGS, and triangle splatting variants under an identical training setup, we provide the first systematic analysis of representation tradeoffs in feedforward scene generation. The project page is available at https://flat-splat.github.io
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。