Runway has introduced a new Gen-3 Alpha model that will be featured across a range of services, with the ability to generate high-fidelity video based on prompts.
An example of a video generated by Gen-3 Alpha. (Image courtesy of Runway)
New Delhi: Runway was a pioneer in the field of generative AI models for creating videos, but was overtaken by OpenAI's Sora and Kuiashou's Kling AI models. Now, Runway is set to return to the fray with the release of its new and improved Gen-3 Alpha model, which has the ability to create photorealistic videos up to 10 seconds long. Sora can generate videos up to 60 seconds long, while Kling can generate videos up to 120 seconds long.
Runway says Gen-3 Alpha is just the first in a series of purpose-trained models on new infrastructure built specifically for training large-scale multimodal models. According to Runway, Gen-3 Alpha represents a significant improvement over Gen-2 in terms of fidelity, consistency and behavior, and is a step toward realizing general-world models: computational models that can simulate, understand and predict a wide range of phenomena.
Gen-3 Alpha has been jointly trained on video and images, enhancing Runway's text-to-video, image-to-video and text-to-image tools with new tools that give you more control over the structure, style and motion of the generated videos, in addition to existing control modes like Motion Brush, Advanced Camera Control and Director Mode.
Gen-3 Alpha can create photorealistic humans. (Image courtesy of Runway)
Photorealistic humans
Runway claims that Gen-3 Alpha excels at generating expressive human characters, capable of generating a wide range of realistic movements, gestures, and emotions. These features are focused on creators, with models that can interpret a wide range of styles and cinematic vocabulary. The models also allow for fine-tuned time control, allowing users to precisely dictate transitions and keyframe elements within a scene.
Technology developed in partnership with industry
Runway revealed that it is working closely and partnering with leading entertainment and media organisations to realise customised versions of Gen-3, which will allow for a more stylistically controlled and consistent character, with a focus on specific artistic and narrative requirements.
