Image composition refers to inserting a foreground object into a background image to obtain a composite image. In this work, we focus on generating plausible shadows for the inserted foreground object to make the composite image more realistic. To supplement the existing small-scale dataset DESOBA, we created a large-scale dataset called RdSOBA with 3D rendering techniques. Specifically, we place a group of 3D objects in the 3D scene, and get the images without or with object shadows using controllable rendering techniques. Dataset is available at https://github.com/bcmi/Rendered-Shadow-Generation-Dataset-RdSOBA.

翻译：图像合成是指将前景物体插入背景图像以生成合成图像。本文聚焦于为插入的前景物体生成合理的阴影，以提升合成图像的真实感。为补充现有小规模数据集DESOBA，我们利用三维渲染技术创建了大规模数据集RdSOBA。具体而言，我们在三维场景中放置一组三维物体，并通过可控渲染技术获取不含或含物体阴影的图像。数据集链接：https://github.com/bcmi/Rendered-Shadow-Generation-Dataset-RdSOBA。