HorizonForge：任意轨迹与任意车辆的驾驶场景编辑
HorizonForge: Driving Scene Editing with Any Trajectories and Any Vehicles
把驾驶场景重建为可编辑的高斯泼溅与网格，再用噪声感知的视频扩散做前馈渲染，为需要精确轨迹/车辆控制的可控视频生成提供了「3D 表示 + 视频扩散先验」的可复用范式。
原文依据 reconstructs scenes as editable Gaussian Splats and Meshes, enabling fine-grained 3D manipulation and language-driven vehicle insertion
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论可控驾驶场景生成对于真实且可扩展的自动驾驶仿真至关重要，然而现有方法难以同时实现照片级真实感与精确控制。我们提出 HorizonForge，一个统一框架，将场景重建为可编辑的高斯泼溅与网格，从而支持细粒度三维操作与语言驱动的车辆插入。编辑结果通过一个噪声感知的视频扩散过程渲染，该过程强制空间与时序一致性，可在单次前馈中产生多样的场景变体，而无需针对每条轨迹进行优化。为规范评测，我们进一步提出 HorizonSuite，一个覆盖自车级与智能体级编辑任务（如轨迹修改与物体操作）的综合基准。大量实验表明，高斯泼溅相比其他三维表示能带来显著更高的保真度，且视频扩散的时序先验对于连贯合成不可或缺。综合这些发现，HorizonSuite 为照片级真实、可控的驾驶仿真建立了一个简单而强大的范式，相较次优的最优方法取得 83.4% 的用户偏好提升与 25.19% 的弗雷歇初始距离改善。
英文原文摘要 1245 字符
Controllable driving scene generation is critical for realistic and scalable autonomous driving simulation, yet existing approaches struggle to jointly achieve photorealism and precise control. We introduce HorizonForge, a unified framework that reconstructs scenes as editable Gaussian Splats and Meshes, enabling fine-grained 3D manipulation and language-driven vehicle insertion. Edits are rendered through a noise-aware video diffusion process that enforces spatial and temporal consistency, producing diverse scene variations in a single feed-forward pass without per-trajectory optimization. To standardize evaluation, we further propose HorizonSuite, a comprehensive benchmark spanning ego- and agent-level editing tasks such as trajectory modifications and object manipulation. Extensive experiments show that Gaussian Splatting delivers substantially higher fidelity than alternative 3D representations, and that temporal priors from video diffusion are essential for coherent synthesis. Combining these finding, HorizonSuite establishes a simple yet powerful paradigm for photorealistic, controllable driving simulation. achieving an 83.4% user-preference gain and a 25.19% FID improvement over the second best state-of-the-art method.