Runway AI has launched Act-One, a feature allowing users to animate AI-generated characters’ facial expressions using simple videos taken with a smartphone. This new tool, which integrates with their existing video generation system, simplifies what used to be a complex task by letting creators map their own facial movements to digital characters without needing advanced equipment.
AI Animation Goes Mobile
Act-One gives creators the power to capture human expressions and bring them to life on AI characters, simply by recording their own faces with an everyday camera. The tool works within Runway’s Gen-3 Alpha model, a system that already supports a variety of video-generation tasks, such as creating clips from text descriptions or images. Runway, with this update, is tackling one of AI-generated video’s weak points: characters with facial expressions that truly match the scene’s tone.
The feature’s rollout began today and is available for users with credits on their accounts. While it’s not free to use, those subscribed to Runway’s video tools can take advantage of this new functionality to craft more realistic videos. An ability to animate micro-expressions and details like eye movements without needing traditional motion capture gear changes the game for small studios and independent creators who don’t have access to the high-end tools used in the film industry.
Gen-3 Alpha Origins
Runway’s Gen-3 Alpha model, which first launched in June 2024, already had the ability to create video clips up to 10 seconds long using various inputs. This model, built with a new infrastructure for multimodal training, set the foundation for the latest updates. It promised high-quality results, emphasizing detailed expressions and smooth transitions between camera angles and scenes.
With a dataset built by experts ranging from scientists to creative professionals, Gen-3 Alpha powers features such as Motion Brush and Advanced Camera Controls, making video generation far more flexible. Since its release, Runway has partnered with several media companies to tailor the model for specific projects, including indie films and other artistic ventures.
Turbo-Powered Expansion
After launching Gen-3 Alpha, Runway kept pushing its capabilities further, introducing the Gen-3 Alpha Turbo model in August 2024, which increased video rendering speed by seven times. This upgrade was critical in allowing faster video outputs without sacrificing quality. The next month, Runway released an API for Turbo, enabling companies to integrate its video generation into their workflows, offering pricing tiers that made it accessible for both individuals and large corporations.
I’m going to say it
This is a game changer!
— Allen T. (@Mr_AllenT) October 22, 2024
The Turbo API debuted in September and quickly attracted interest from industries outside of traditional video production, including marketing agencies and digital content creators. Omnicom, a global marketing company, is among the firms already using this API, and early partners are offering positive feedback, noting how the API’s speed and ease of use have made integrating AI video tools more efficient.
Making Motion Capture Accessible
Before Act-One, generating realistic facial animations required complex setups, usually involving motion-capture suits, specialized cameras, and meticulous editing. Films like Avatar and Planet of the Apes employed these methods, which are expensive and time-consuming.
By comparison, Runway’s tool allows for a much simpler setup—just a single video input can produce results similar to what you would see in Hollywood productions. This opens up more opportunities for independent filmmakers, game designers, and content creators working on a tighter budget.
Notably, Runway’s Act-One feature lets creators animate characters in multiple styles and designs, keeping the expressions natural and fluid. For those working in storytelling, this means characters can display emotion across different animation styles, from hyper-realistic to stylized, all with high fidelity to the source performance.
Partnerships and Ethical Considerations
Runway has gained traction in the entertainment industry, partnering with Lionsgate for custom AI video models based on their catalog of over 20,000 titles. These partnerships allow major studios to use AI tools for large-scale projects without needing to rely on traditional, more time-consuming methods of character animation.
Despite its rapid rise, Runway has had to navigate some of the same ethical challenges as other AI firms. The company has been cautious about the data it uses to train its models, but concerns about intellectual property remain prevalent across the industry.
In July, a report from 404 Media claimed Runway’s internal documents listed YouTube channels from major companies like Netflix and Disney as data sources. Although the company hasn’t disclosed all its training data, Runway maintains that its models are built with internally curated datasets.
Runway is also introducing several safeguards to prevent Act-One from being misused. Public figure impersonations, for instance, are blocked, and voice usage rights are verified when using this tool in combination with other features.
A Competitive AI Video Market
Runway’s quick expansion in 2024 has placed it in direct competition with major players like OpenAI, Adobe, and Google, who are all exploring video generation powered by AI. OpenAI’s Sora, expected by the end of 2024, is anticipated to be a major rival in this space, while other companies like Luma AI are focusing on similar technologies, particularly around camera control within AI-generated environments.
Despite this, Runway’s focus on making high-quality video generation accessible to smaller creators sets it apart from these larger players. With a valuation of $1.5 billion and backing from prominent names such as Salesforce and Google, the company has secured a strong position in the rapidly evolving generative AI landscape.
Last Updated on November 7, 2024 2:23 pm CET
