Text-to-3D
Text-to-3D is the task of generating a 3D object or scene from a written prompt. Image-to-3D does the same from one or more photos. The output can be viewed from any angle.
Also called: image-to-3D, 3D generation, AI 3D model generator
By Divya Sharma · Checked Nov 15, 2026 · How we review
Where text-to-image produces a flat picture, text-to-3D produces something you can rotate, light and place in a game engine or AR scene. Output formats include meshes, neural radiance fields (NeRFs) and Gaussian splats.
How it works
Good 3D training data is scarce compared with images, so early methods borrowed from 2D models. Google's DreamFusion, from 2022, optimised a NeRF so that renders of it from random angles looked right to a pretrained 2D image diffusion model. It needed no 3D training data, but each object took a long optimisation run. Newer systems are trained on 3D datasets or generate several consistent views first and then reconstruct the shape, which is much faster. Image-to-3D works the same way, starting from a photo instead of a prompt.
Uses
Game props, product previews, AR filters, rapid prototyping and concept art. Many tools export a mesh you can clean up in Blender. Related work on generating whole explorable scenes is covered under world model.
Limits
Output often needs cleanup before production: messy topology, baked-in lighting, blurry textures and the occasional object with a face on both sides. Fine mechanical detail and exact dimensions are weak spots, so it is better for drafts than for parts that must fit.
Related terms
- Text-to-image
- Text-to-image is the task of generating a new picture from a written description, called a prompt. Most current text-to-image models are diffusion models conditioned on the text.
- Diffusion model
- A diffusion model is a generative model that learns to turn random noise into data, such as an image, by removing a little noise at a time over many steps.
- Image-to-image
- Image-to-image is the task of generating a new image from an existing one plus a prompt, keeping some of the original's structure while changing style, detail or content.
- World model
- A world model is an AI model that learns how an environment works well enough to predict what happens next, often from video or sensor data, so an agent can plan or train inside the simulation.
- Multimodal model
- A multimodal model is an AI model that can take in, and sometimes produce, more than one kind of data, such as text, images, audio or video, within a single model rather than through separate systems.