Create Realistic Video Content with Google AI and Runway: Text Prompts for Automatic Video Editing

Recently, Google entered a significant partnership with Runway, providing millions in cloud services to enhance artificial intelligence advancements in film production. Runway specializes in using AI to create surreal video content from text descriptions. Their technology notably contributed to the acclaimed sci-fi film "Everything Everywhere All At Once," where iconic scenes, such as the noodle hands and stone dialogues, were crafted using Runway’s tools.

The film's success has drawn investment from leading firms like Amplify Partners and Lux Capital. Runway has now launched its new AI-powered video editing tool, Gen-2, allowing users to generate "realistic video content" and "automatically edit videos" with simple text prompts.

Initially unveiled in February, Runway's Gen-1 model enabled users to alter video styles through text descriptions, transforming real-life footage into animated art. Gen-2 advances this by enabling scene creation from a single text prompt; for example, entering "drone shot of mountains" can produce a realistic aerial view. While the current open-source model yields lower resolution outputs and requires further development for improved quality, it can also generate short video sequences from combinations of images and text.

Currently, Gen-2 is available to testers registered on Discord, with Runway stating that Gen-1 already serves "thousands of users." The company's vision is to blend human creativity with AI-generated content while mitigating potential misuse of artificial intelligence. Gen-2 is accessible on the Apple App Store and via web browsers.

While Gen-2 boasts enhanced features, it is priced higher, with the most affordable package costing nearly a thousand yuan per year. Users may face limitations, including slow generation speeds—around three minutes for a five-second video—and occasional abstract video outputs.