DreamFusion is a product that utilizes a text-to-image generative model named Imagen to enhance a 3D scene.

This product offers Score Distillation Sampling (SDS), which is a method for generating samples from a diffusion model by optimizing a loss function. SDS allows for the optimization of samples in any parameter space, including a 3D space.

To define this differentiable mapping, the product adopts a 3D scene parameterization similar to Neural Radiance Fields. In addition, DreamFusion incorporates supplementary regularizers and optimization strategies to enhance geometry. This results in the production of coherent and high-quality normals, surface geometry and depth, as well as relightable NeRFs featuring a Lambertian shading model.

The applications for DreamFusion involve optimizing 3D scenes for various purposes, such as video games, virtual reality experiences, and film production.