KwaiVGI/CameraClone-Dataset

Camera Clone Dataset是一个为相机克隆学习设计的大规模合成数据集，包含多样化的场景、主体和相机移动。该数据集由三个视频组成的视频集构成，包括相机运动参考视频、内容参考视频和目标视频。使用Unreal Engine 5进行渲染，涵盖了40个3D场景，66个人物模型和93种动画，以满足不同场景的需求。相机轨迹类型包括基本移动、弧形路径和随机路径，数据集总计有391K个视频和1,155K个三重视频集。

The Camera Clone Dataset is a large-scale synthetic dataset designed for camera clone learning, featuring diverse scenes, subjects, and camera movements. It consists of video sets with three videos each: a camera motion reference video, a content reference video, and a target video that recaptures the scene in the content reference video with the same camera movement as the camera motion reference video. Rendered using Unreal Engine 5, the dataset includes 40 3D scene backgrounds, 66 character models, and 93 different animations. Camera trajectories are categorized into basic movements, arc paths, and random paths, totaling 391K videos and 1,155K triple video sets.