DeepMotion, a leading AI motion-capture specialist, has recently unveiled SayMotion, a cutting-edge browser-based generative AI service designed to transform text prompts into captivating character animations. This innovative platform, currently available for free during its open beta phase, produces animations in FBX, GLB, and BVH formats, tailored for integration into DCC software and game engines.
SayMotion sets out to offer users an intuitive method for creating foundational motions suitable for a wide array of applications, including games, AR, VR, animation, and motion graphics projects. Founded in 2014 by industry veteran Kevin He, DeepMotion has long been at the forefront of AI motion capture technology, setting the stage for the emergence of various browser-based mocap tools.
With SayMotion, DeepMotion ventures into the realm of generative AI, empowering users to craft animation clips by inputting text descriptions of desired movements. Prompts are structured into three parts: subject, action, and detail. The subject defines the character type (e.g., ‘woman’, ‘child’, ‘swimmer’), while the action describes the specific movement being performed. The detail component allows for further customization by indicating direction, orientation, style, or emotion.
By stringing together these prompts, users can generate intricate sequences of actions. For instance, a prompt like ‘A person walks clockwise in an angry manner, reaches out with their left arm to grab the left arm of a chair, and lowers themselves heavily into the seat’ showcases the potential complexity achievable through SayMotion.
Moreover, SayMotion offers an ‘inpainting’ feature that enables users to refine initial animations by extending or blending them with additional text prompts. This functionality allows for the enhancement and modification of animations, as demonstrated in the starter guide where actions like bending down to pick up an object are seamlessly integrated into existing animations.
In addition to supporting DeepMotion’s pre-made characters and built-in avatar-generation tools, SayMotion accommodates custom 3D characters uploaded by users. Whether utilizing DeepMotion’s assets or importing custom models from platforms like Mixamo or Sketchfab, animators can export their creations in FBX, GLB, or BVH formats for seamless integration into popular software applications and game engines.
While SayMotion’s current focus is on generating full-body motion for human characters, DeepMotion is exploring the possibility of incorporating facial and hand animation capabilities in the future. The company also plans to introduce video prompts, enabling the creation of unique animations based on SayMotion’s interpretation of source videos.
During the open beta phase, users receive 2,000 credits per month to utilize for generating animations, downloading content, or utilizing the inpainting feature. Animations created during this period are licensed for commercial use. Although DeepMotion has not disclosed final pricing or release dates for SayMotion, interested parties can explore Animate3D subscription plans for further details.
For a comprehensive overview of how SayMotion operates, visit DeepMotion’s blog and product webpage. Stay updated on the latest developments by following CG Channel on social media platforms like Facebook and Instagram. Join the conversation and gain access to exclusive content showcasing behind-the-scenes insights into VFX movies, animations, game cinematics, and motion graphics projects.
