Can A.I Make Videos? - Methods in Development

A.I videos production, and text-2-video is coming, and it might be closer than you think.

Art generated by artificial intelligence has been taking the internet by storm recently. The most popular tools such as DALL-E 2, Midjourney, and Stable-Diffusion, are all pretty good at generating artwork from text input. Related: We made some art using A.I.

Some methods to blend between images can be used to create moving illustrations. These type of creations tend to have a dreamlike quality, and it can be difficult to create coherent scenes/takes. 

Check out this music video made with A.I.

What's missing, photorealism? Stable-Diffusion can get pretty photorealistic from what I've seen. The thing is, these A.I are trained on pictures and art, not videos; hence they don't have much data on how things move. 

Meta shows off their Make-A-Video app which was trained on both images and videos, and can do Text-to-video with some impressive results.

"The system uses images with descriptions to learn what the world looks like and how it is often described. It also uses unlabeled videos to learn how the world moves. With this data, Make-A-Video lets you bring your imagination to life by generating whimsical, one-of-a-kind videos with just a few words or lines of text."

A.I for video editing is already a thing in apps like Runway, where you can do tracking, background removal, and some other effects powered by machine learning.

To create animated scenes, a different approach might be necessary. Maybe involving 3D modeling and character rigging, like this A.I that learned to animate humanoids.