Text-to-video
Text-to-video is the task of generating a short video clip from a written prompt. Models must produce frames that each look right and also move consistently from one frame to the next.
Also called: video generation, T2V, AI video generator
By Divya Sharma · Checked Nov 15, 2026 · How we review
Text-to-video extends text-to-image with a second problem: time. A single good frame is not enough. Objects must keep their shape, lighting must stay stable and motion must look physical across every frame.
How it works
Many video models start from an image diffusion model and add layers that look across frames. Meta's Make-A-Video paper put the idea plainly: learn appearance from paired text and images, and learn motion from unlabelled video. Stable Video Diffusion trained in stages, first on images, then on a large video set, then on a smaller high-quality one.
In practice
Hosted services such as Kling and Runway, and models such as Veo 3.1, produce clips of a few seconds from a prompt. Many also accept a still image as the first frame, which is image-to-video. More options are listed under video models.
Limits
Clips are short, and longer scenes are usually stitched together. Fine detail, hands, text and cause and effect often break down. Generation is far more expensive than images, since every second holds many frames. Realistic output also raises deepfake concerns.
Related terms
- Image-to-video
- Image-to-video is the task of turning a still image into a short video clip, usually with the image as the first frame and a text prompt describing the motion.
- 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.
- 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.
- Lip sync (AI video)
- AI lip sync is the use of a model to change the mouth movements of a person in a video or image so they match a new audio track, such as a translated or rewritten line.
- Deepfake
- A deepfake is image, audio or video content made or altered with AI so that it convincingly shows a real person, place or event in a way that did not happen and could pass as authentic.
- 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.