DisCoScene: Spatially Disentangled Generative Radiance Fields for Controllable 3D-aware Scene Synthesis.
Yinghao XuMenglei ChaiZifan ShiSida PengIvan SkorokhodovAliaksandr SiarohinCeyuan YangYujun ShenHsin-Ying LeeBolei ZhouSergey TulyakovPublished in: CoRR (2022)
Keyphrases
- global illumination
- video sequences
- d scene
- three dimensional
- generative model
- complex scenes
- single image
- photometric properties
- scene understanding
- multiple images
- computer vision
- dynamic scenes
- image sequences
- camera response function
- input image
- real scenes
- stereo reconstruction
- viewpoint
- spatial location
- outdoor scenes
- plenoptic function
- moving objects
- scene geometry
- discriminative learning
- image based rendering
- lighting conditions
- probabilistic model
- geometric constraints
- multiple objects
- illumination conditions
- object detection