Artificial intelligence startup Runway launched a text-generated video model with excellent results.

CTOnews.com, March 20 (Xinhua)-- an artificial intelligence startup called Runway has announced a new text-generated video (Text-to-video) AI model called Gen-2, in which users can enter a description and the system automatically generates any style of corresponding video. Current technology doesn't live up to this dream, but Runway's new model has shown some very good results.

Runway provides a web-based video editor that focuses on artificial intelligence tools such as background removal and pose detection. The company helped develop the open source text generation image model Stable Diffusion and released the first artificial intelligence video editing model Gen-1 in February.

The main purpose of Gen-1 is to convert existing video material, allowing users to enter a rough 3D animation or jitter shot on the phone, and apply an artificial intelligence-generated overlay. For example, in the following example, a video of cardboard packaging is combined with a picture of an industrial plant to produce a clip that can be used for storyboards or proposals for more beautiful works.

By contrast, Gen-2 is more focused on generating video from scratch, but there are a lot of things to pay attention to. First of all, the demos shared by Runway are short, unstable and not very realistic; second, access is limited, and users must register through Runway's Discord platform and join the waiting list to use the Gen-2 model; "We will provide extensive access in the coming weeks," said Kelsey Rondenet, a spokesman for the company. In other words: currently we can only evaluate the Gen-2 model through a demo and a few clips (most of which have been promoted as Gen-1 models), but the results seem to be excellent.

Text generation video technology is exciting and brings new creative opportunities, but it also brings new threats (such as false information, etc.). CTOnews.com will keep an eye on Runway's new technology.