Learning Object-Centric Neural Scattering Functions for Free-viewpoint Relighting and Scene Composition.
Hong-Xing YuMichelle GuoAlireza FathiYen-Yu ChangEric Ryan ChanRuohan GaoThomas A. FunkhouserJiajun WuPublished in: Trans. Mach. Learn. Res. (2023)
Keyphrases
- learning objects
- free viewpoint
- scene geometry
- real scenes
- d scene
- super resolved
- multi view
- depth map
- e learning
- lighting conditions
- cast shadows
- metadata
- view synthesis
- depth estimation
- learning process
- learning object repositories
- super resolution
- virtual view
- three dimensional
- single image
- reference images
- learning object metadata
- high resolution
- dynamic scenes
- low resolution
- stereo vision
- depth information
- complex scenes
- image sequences
- image synthesis
- input image
- video sequences
- illumination conditions
- geometric information
- light field
- computer graphics
- depth images
- scene understanding
- high quality
- databases