Runway says its latest AI video model can actually generate consistent scenes and people

# ai # News # Tech

Runway, an AI startup, has just announced that their latest AI video model, Gen-4, can keep scenes and characters consistent across different shots. This is a big deal because AI-generated videos often have trouble with maintaining a smooth storyline. But according to Runway's post on X, Gen-4 is set to give users more "continuity and control" when they're crafting their stories. Right now, Gen-4 is being rolled out to folks who pay for the service and enterprise users. This new model lets you create characters and objects that stay the same across various shots, all starting from just one reference image. All you gotta do is describe what you want the scene to look like, and Gen-4 will whip up consistent outputs from different angles. To show off what Gen-4 can do, Runway dropped a video of a woman who looks the same in different shots, even when the lighting changes and the context shifts. This release comes hot on the heels of their Gen-3 Alpha video generator, which they announced less than a year ago. Gen-3 let users make longer videos, but it stirred up some controversy because it was reportedly trained on a bunch of YouTube videos and pirated movies that were scraped from the internet.

Related article

Comments (36)