MVNT AI
From Motion to Meaning: Understanding and Synthesizing Dance
At MVNT, we combine our technical expertise with a deep connection to dance, working to address the real challenges faced by the global dance community. As a dance IP infrastructure company, we create tools to protect, enhance, and share choreography in meaningful ways. As a collective of engineers, architects, and artists ourselves—we are deeply aligned with responsibilities to pioneer & protect the creators' rights and thus contributing to develop an AI ecosystem safeguarding choreographic works to enable new creative possibilities of movements in the virtual worlds.
The future of the entertainment industry lies in immersive and interactive experiences along with the growing demand for human-like features of avatars and characters. In addition to increasing adoption of AR/VR, there are growing demands of high-fidelity 3D animation motions across the production side of film, television, video games, animation, robotics, and sports science.
AI labs within big tech giants such as NVIDIA, Meta, Epic Games, Tencent all share the common goal of creating better text-to-motion or audio-to-dance generative AI tools & methodologies for enhanced development procedures and cross-industrial implementations. Some remarkable usecases include:
Games: The concept of CPC (Co-Playing Character), which is scheduled to be introduced in Krafton's inZOI (and possibly GTA6) implementing solutions in NVIDIA's ACE Program
Music: Demand for high-quality animation asset production and IP operation is increasing worldwide with digital human performances such as Anyma (DJ), naevis, PLAVE (K-Pop, VTuber) and more.
However, multiple challenges remain when it comes to any type of movement-related AI—variability in motion, synthesis fidelity under prompt conditioning, temporal coherency and more. Such quality limitations occur due to the lack of sophisticated data on movement and the lack of professional model design for dance/motion understanding.
At MVNT, we are given competitive advantage such as:
collecting multiple dance motion genres including latest K-Pop choreographies, hip-hop, ballet, and other dances around the world;
state-of-the-art motion capture production studio (using Vicon Vantage / Optitrack Prime) acknowledged by our partners such as Naver Z and Krafton;
dataset and model architecture in collaboration with dance and AI experts, with analysis of fine-grained movement classification, annotations, and music as our conditioning input.
Our work centers on two pillars:
Dance Recognition creates a foundation for understanding choreography. It supports tools for IP management, monetization, onboarding, and trend analysis—and will further enable features like TOP-100 trending charts and skill assessment tools.
Dance Generation offers AI tools for game developers, entertainment studios, and choreographers to create, refine, and customize choreography. Integrated into the MVNT STUDIO creative suite, it helps bring dance into virtual spaces and interactive media.
We want to streamline the overall workflow of producing high-quality dance data for producers through our datasets and models. By focusing on these two pillars, MVNT is building the infrastructure and capabilities to expand the dance industry and establish itself as the leading technology company for dancers worldwide.
Own your movement.
MVNT
Last updated
