Runway’s mission is to create expressive and controllable tools for artists that open new possibilities for creative expression. Today, Runway introduces Act-One, a cutting-edge tool for generating dynamic character performances within Gen-3 Alpha. Act-One enables the creation of compelling animations using both video and voice performances as inputs, representing a significant advancement in the use of generative models for expressive live-action and animated content.
Capturing the Essence of a Performance
Traditional facial animation workflows often rely on complex, multi-step processes. These typically involve motion capture equipment, multiple reference footage sources, manual face rigging, and other techniques, all aimed at transferring an actor’s performance into a 3D model for animation. The primary challenge with these methods is maintaining the emotion and subtle nuances of the original performance when translating it to a digital character.
Runway’s approach uses a completely different pipeline, driven solely by the actor’s performance and requiring no additional equipment.
Animation Mocap
Act-One by Runway is capable of working with a wide range of reference images, preserving realistic facial expressions while accurately translating performances into characters with proportions that differ from the original source video. This flexibility enables new creative possibilities for character design and animation, offering greater freedom in crafting unique and imaginative characters.
Live Action
Runway Act One also excels at producing cinematic, realistic outputs, demonstrating remarkable robustness across various camera angles while maintaining high-fidelity facial animations. This ability enables creators to craft believable characters that convey authentic emotion and expression, deepening the viewer’s connection to the content.
New Creative Avenues
Runway has been exploring how Act-One can facilitate the generation of multi-turn, expressive dialogue scenes—something that has been difficult to achieve with generative video models in the past. Now, creators can produce narrative content using just a consumer-grade camera and a single actor performing multiple characters from a script, significantly streamlining the process of bringing complex scenes to life.
Safety
As with all of Runway’s releases, responsible development and deployment remain a top priority. Act-One is being launched with a comprehensive set of content moderation and safety precautions, which include:
- Enhanced capabilities to detect and block attempts to generate content featuring public figures;
- Technical measures to ensure that users have the right to use voices created with Custom Voices;
- Continuous monitoring to identify and mitigate potential misuse of the tools and platform.
These safeguards are part of Runway’s broader “Foundations for Safe Generative Media,” which guide the responsible release of Act-One and future innovations.
Runway Act-One: Looking Ahead
Runway is excited to see the new forms of creative storytelling that Act-One will inspire in animation and character performance. This tool represents another significant step toward making advanced techniques accessible to a wider range of creators and artists.
The team looks forward to seeing how artists and storytellers use Act-One to bring their visions to life in innovative and exciting ways. Access to Act-One will begin gradually rolling out today and will soon be available to all users.
Coming soon at https://runwayml.com/
Explore other Gen-3 capabilities in our Blog.