MAV3D (Make-A-Video3D) is a method for generating three-dimensional dynamic scenes from text descriptions. Our method employs a 4D dynamic Neural Radiance Field (NeRF) that is optimized for scene appearance, density, and motion consistency by querying a Text-to-Video (T2V) diffusion-based model. The dynamic video output generated from the provided text can be viewed from any cameraContinue reading “Dynamic Text-to-4D Scene Generation”
-
Follow
Following
Already have a WordPress.com account? Log in now.
Random 101 