Meta presented with MAV3D (Make-A-Video-3D) a method for generating 4D content, i.e. a 3D video, from a text description by using a 4D dynamic Neural Radiance Field (NeRF) [project page, paper]. Unfortunately, the source code has not been released.