Artificially intelligent (AI) text-to-image generators have been lighting up the airways so far this year but now a company has teased an AI text-to-video generator.
Video editing software company Runway put out an awesome promotional video where it teases creating any scene that you can think of but as a video.
The video starts with an operator typing in “Import city street.” A street scene is created and then tweaked with the input “Make it look more cinematic.”
Perhaps most impressively, is when a lamppost is removed from the moving scene by using a type of spot healing tool.
The video then switches to “Generate a lush garden.” Once entered, a number of different scenes are spawned for the user’s perusal that includes many different styles.
After choosing a scene, the user can choose an animation option such as scale in overtime, as well as create text to fade in and out.
Next, a video of a photographer appears and “Green screen character” is inputted into the prompt box. This separates the subject from the background, allowing the operator to blur it and then make it black-and-white.
Runway has been a video editor for some time, but this new AI offering appears to be totally novel in the film-cutting space. The software does already have AI features, such as the ability to mask objects in a video.
AI Video Generators Face Difficulties
As of today, the training models for AI video are primitive and there are also issues because of video’s high computational demands. However, this time last year AI-image generators were in a similar state so it is not a stretch to imagine that AI-video generators may come on leaps and bounds in the coming months.
Runway is a browser-based app and is available now for a monthly fee. But the advertised text-to-video feature is still currently in an early access phase and to test it out users must sign up for the waitlist on Runway’s website.