Alignment Is All You Need For X-to-4D Generation

Generative diffusion models excel at synthesizing high-quality images, videos, and 3D content under multimodal control.

Generative diffusion models excel at synthesizing high-quality images, videos, and 3D content under multimodal control. However, arbitrary user-defined modality-to-4D (X-to-4D) generation remains challenging due to the high cost of constructing diverse datasets and the limited…