4 mins read

Meta’s Emu Video Will Transforms Any Text into a Smooth Video

Big news from Meta: they’ve just shared two research results focused on Generative AI: Emu Video and Emu Edit. Let’s dive into what these cool tools are all about.

Emu Video: Making Videos with Just a Few Words

Ever thought you could make a smooth video by just describing it? That’s what Emu Video does.
Emu Video, powered by Meta’s Emu model, offers a straightforward method for creating videos from text, using advanced diffusion models. It simplifies video generation into two main steps: creating an image from text, then crafting a video from both. This method, which streamlines the traditional multi-model approach down to just two models, efficiently produces high-quality 512×512 videos at 16 fps. It has been highly favoured in user tests for its quality and loyalty to the original text prompt, and it also excels in animating images with text instructions, setting a new industry standard.

Why’s Emu Video Special?

Unlike other tools, Emu Video keeps it simple. It uses fewer steps to make these videos, which means it’s quicker and easier. People who tested it really liked it more than other video-making AI. They said it matched their ideas better and looked great too.

emu video examples
Example videos variations from Emu Video

Emu Edit: Changing Photos with Just a Sentence

Emu Edit represents a significant advance in generative AI, specifically tailored for editing images with a high degree of precision. It understands the nuances of image manipulation, accommodating users’ desires for specific changes without altering the whole picture. This tool is particularly adept at detailed tasks: whether you’re aiming to subtly adjust the backdrop of a photo or to boldly transform objects within it. Emu Edit’s capabilities come from a vast training dataset of 10 million samples, which grants it the ability to follow complex instructions and deliver edits that are true to the user’s intent. This results in not just believable, but also accurately edited images, maintaining the quality and specificity requested by the user. With Emu Edit, Meta aims to redefine the standards for AI in image editing, providing an unparalleled level of control and quality in the final edited images. Hopefully we will be able to use this model soon!

Explore and Play: The Emu Video Demo Page

Meta has put up a demo page where you can play with Emu Video. It lets you choose phrases to create your own video. Pick from cute phrases like “a fawn Pembroke Welsh Corgi” or “a panda wearing sunglasses,” mix it up with actions and places; the results look smooth and impressive. It’s a playground for your imagination and a sneak peek into what you could do with Emu Video in the future.

meta emu video demo page

What’s Next?

Right now, Emu Video and Emu Edit are like a sneak peek into the future. They’re not ready for us to be used yet, but they show us what might be possible soon. Imagine making your own fun GIFs or creating videos from a simple text prompt. Well, technically we can already do it now, with solutions like AnimateDiff and Stable Diffusion, but the problem of consistency remains, unless other tools like ControlNet are used. What Meta revealed could be a game changer in this aspect.


Meta’s new tools are about making AI fun and easy for everyone. Whether you’re an artist looking for new ideas or just want to make cool stuff for your social media, Emu Video and Emu Edit are like a glimpse into a future where we can all be creators. And that’s pretty exciting, right? For now, it’s just research, but we are getting closer to an easy and functional text-to-video generation.

Meanwhile, if you’d like to generate AI video on your PC, discover Stable Video Diffusion.


Leave a Reply

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.