advertise image

Meta launches tool that generates videos from text

Meta announced Make-A-Video on Thursday. It is an AI-powered video generator that can create novel video content from text or image prompts. It is similar to existing image synthesis tools like DALL-E and Stable Diffusion. It is also able to make variations of existing videos. However, it is not yet available for public use.

On its announcement page, Meta had shown example videos generated from text, including a young couple walking in heavy rain and a teddy bear painting a portrait. It also showcases Make A Videos’ ability to take a static source image and animate it. For example, even a still photo of a sea turtle can appear like swimming after processing through the AI model. There is a crucial technology behind Make-A-Video. Its builds off existing work with text-to-image synthesis used with image generators like Open AI’s DALL-E. In July, Meta announced their own text–to–image AI model called Make-A-Scene.

Meta has not trained the Make-A-Video model on labeled video data. It had captioned descriptions of the actions depicted. Meta had instead taken image synthesis data where images are still prepared with captions and had applied unlabelled video training data so that the model learns a sense of where a text or image prompt might exists in time and space. It can then predict what comes after the image and display the scene in motion for a short period. AI research has played a vital role in pushing creative expressions forward. They have done this

s by providing users with easy tools to create new content. With just a few sentences or words, Make-A-Video can bring users’ imagination to life and generate one-of-a-kind videos with vivid colors, landscapes, and characters.
The Artificial Intelligence system can create videos from images and existing videos to create new and similar content. Meta is continuously committed to open science. They shared the details of the system in a research paper. They are also planning to release a demo experience for their users.

Featured Image Credit: Techzine Europe

Leave a Reply

Your email address will not be published. Required fields are marked *