LightillusionsLab/SceneMaker

SceneMaker 是一个基于 Objaverse 的 3D 场景渲染数据集。该数据集为每个场景提供了多视角的渲染图像，包括 RGB、深度图和分割图。每个场景（由 UUID 标识）包含主体物体（来自 Objaverse 的原始物体）、四个随机选择的额外物体、地板和环境贴图。每个场景包含 20 个不同的视角，每个视角提供 RGB 彩色渲染图像、深度图和分割图（语义分割或实例分割）。SceneMaker is a 3D scene rendering dataset based on Objaverse. The dataset provides multi-view rendered images for each scene, including RGB, depth maps, and segmentation maps. Each scene (identified by UUID) contains a main object (original object from Objaverse), four randomly selected additional objects, a ground plane, and an environment map. Each scene includes 20 different views, with each view providing RGB rendered images, depth maps, and segmentation maps (semantic or instance segmentation).