Blender-Color-Edit

该数据集是由Blender构建的，包含了五个3D物体，每个物体都制作了两种不同的外观，并涉及到了经典的摄像机运动，如平移和缩放。此外，该数据集为颜色化和一般外观编辑等任务提供了确定性的评估指标，并使用辅助指标来评估时间一致性。规模上，该数据集包含了五个具有多种外观的3D物体，其任务重点是参考基础的视频外观编辑。

This dataset was constructed using Blender. It consists of five 3D objects, each with two distinct appearance variants, and incorporates classic camera motions including translation and zooming. Furthermore, this dataset provides deterministic evaluation metrics for tasks such as colorization and general appearance editing, alongside auxiliary metrics for assessing temporal consistency. In terms of scale, this dataset includes five 3D objects with diverse appearances, and its task focuses on reference-based video appearance editing.