D-NeRF (Dynamic Neural Radiance Fields) is a novel method for synthesizing novel views of dynamic scenes with complex, non-rigid geometries. Unlike traditional methods that require multiple views or ground-truth geometry, D-NeRF uses a single monocular camera and trains an underlying deformable volumetric function from sparse input monocular views. The method decomposes the learning process into two stages: encoding the scene into a canonical space and mapping this representation to the deformed scene at a specific time. Both mappings are learned using fully-connected networks. D-NeRF can control both the camera view and time variable, allowing it to render high-quality images of objects under rigid and non-rigid motions. The effectiveness of D-NeRF is demonstrated through experiments on scenes with various types of deformations, including articulated motion and human body poses. The method also produces complete 3D meshes that capture time-varying geometry, all from a single viewpoint.D-NeRF (Dynamic Neural Radiance Fields) is a novel method for synthesizing novel views of dynamic scenes with complex, non-rigid geometries. Unlike traditional methods that require multiple views or ground-truth geometry, D-NeRF uses a single monocular camera and trains an underlying deformable volumetric function from sparse input monocular views. The method decomposes the learning process into two stages: encoding the scene into a canonical space and mapping this representation to the deformed scene at a specific time. Both mappings are learned using fully-connected networks. D-NeRF can control both the camera view and time variable, allowing it to render high-quality images of objects under rigid and non-rigid motions. The effectiveness of D-NeRF is demonstrated through experiments on scenes with various types of deformations, including articulated motion and human body poses. The method also produces complete 3D meshes that capture time-varying geometry, all from a single viewpoint.