Text To 3D

lastmod 2026-11-10created 2023-04-065 edits

Text To 3D is a deep learning task that generates 3D models (meshes, NeRFs, Gaussian splats, etc.) of objects or scenes described in natural language sentences. Because 3D training data is scarce, a widely used approach, as in DreamFusion, is to optimize a 3D representation by distilling the knowledge of a pretrained 2D text-to-image diffusion model from multiple viewpoints (score distillation), and many follow-up studies have since improved speed and quality. This page collects related projects along with pages on adjacent generation tasks such as text-to-image and text-to-video.

Projects

• DreamFusion - Text-to-3D using 2D Diffusion

See also

• Deep learning
• Text To Image
• Text To Video
• Text To 3D
• Text To Music