Vision transformer empowered physics-driven deep learning for omnidirectional three-dimensional holography | Synapse