"KAIST Geeks Did It" — MOVIN Releases Ultra-Affordable Motion Capture

[Startup Story] Byeoli Choi, CEO of MOVIN

In Avatar, a film that revolutionized cinematic paradigms, what stands out as much as its dazzling backdrops is the characters’ movement. The blue-skinned aliens’ lifelike expressions and gestures are brought to life through motion capture technology, which recreates human performances in 3D on screen.

Once the domain of major studios, motion capture (mocap) is now rapidly expanding—particularly in the rise of virtual artists, or online streamers who perform through digital avatars. Thanks to improved accessibility, anyone can become a virtual performer from home.

Yet, reproducing detailed, real-time motion like Avatar still poses challenges. Common issues such as lag, delayed recognition, or characters flying off-screen due to tracking errors remain unsolved. But MOVIN, a Korean startup, has developed tech that captures motion with Avatar-level fidelity—without any body-worn markers.

Breaking Cost and Space Barriers: Mocap, Reimagined by KAIST Enthusiasts

“We’re self-professed geeks,” said Byeoli Choi, CEO of MOVIN. “Back in 2022, our founding team at KAIST was fascinated by the virtual idol project ISEGYE IDOL. The idea of 3D avatars interacting with fans in a virtual space was inspiring—but the technical limitations were clear.”

MOVIN was founded in August 2023 by computer graphics researchers at KAIST’s Graduate School of Culture Technology. The team—previously part of the LAVA Lab (Lifelike Avatar and Agents)—had been researching how to make avatar movements more realistic using AI-based deep learning.

Mocap technology traditionally falls into three categories: optical, inertial, and AI-based. Optical systems, like those used in Avatar, require 20–30 infrared cameras to track body-mounted markers. Inertial systems use gyroscopes attached to the body to measure motion through angles and acceleration.

Each has trade-offs. Optical systems offer high precision but are extremely expensive, often costing billions of KRW (millions of USD) per studio setup. Inertial systems are cheaper but less accurate, especially during intense movements, due to gyro drift.

Choi explained, “Both systems involve wearing cumbersome gear—markers or sensors—that limit natural movement. If they shift during action, the system fails to track properly, and you end up with time-consuming post-processing.”

LiDAR-Based AI Mocap: Compact, Scalable, Accurate

To overcome these limitations, MOVIN developed AI-based motion capture. “Our system uses AI to interpret motion directly from camera input—no markers or wearable gear required,” said Choi.

However, conventional AI mocap systems still rely on 2D camera input, which can’t fully capture spatial depth. “To track 3D motion, you'd need multiple cameras, and stitching that data together is both technically difficult and expensive,” Choi noted.

MOVIN solved this using LiDAR (Light Detection and Ranging). “LiDAR uses laser pulses to calculate precise 3D position data,” said Choi. “We feed this spatial data into our AI models to recreate full-body 3D motion—using just one LiDAR sensor.”

Their flagship product, MOVIN TRACIN, delivers approximately 90% of the accuracy of optical systems at just 1% of the cost. Plus, it’s fully portable and works indoors or outdoors. Applications go far beyond content creation—Choi envisions use cases in sports, physical rehabilitation, and construction site safety diagnostics, anywhere high-quality 3D motion analysis is needed.

MOVIN’s technology has already earned recognition in the global market. At GDC 2024, the world’s largest game development conference, MOVIN’s prototype received an enthusiastic reception. Companies like Meta, Microsoft Xbox Studios, and HTC VIVE placed over 40 preorders across 19 global buyers.

“MOVIN is not just a hardware company,” Choi emphasized. “Our goal is to become a 3D motion AI platform—democratizing access to high-quality motion data for creators, developers, and researchers around the world.”