DeepMotion provides a cloud‑based AI platform that converts standard video into rigged 3D skeletal and facial animation data. Users upload clips via a web portal or API and receive assets in FBX, BVH, or glTF, with automatic retargeting to common rigs for Unity, Unreal, and other pipelines. The service eliminates the need for physical mocap hardware, offering fast, scalable processing for indie developers, XR creators, and visual‑effects studios.
Funding
Funding not disclosed
Founders
Product
Problem
Traditional motion capture workflows rely on expensive camera rigs, marker sets, and dedicated studio space, which creates high entry costs and long turnaround times for realistic 3D character animation. Small studios, indie developers, and XR creators often lack the resources or expertise to operate such systems, limiting their ability to produce high‑fidelity digital humans. Consequently, production pipelines become fragmented and costly, slowing iteration cycles.
Solution
DeepMotion offers an AI‑powered motion capture and body‑tracking platform that transforms ordinary video footage into fully rigged 3D skeletal and facial animation data. The service runs on a cloud‑based inference engine that applies deep neural networks for pose estimation, mesh reconstruction, and facial expression mapping, eliminating the need for physical mocap hardware. Users upload video clips through a web portal or via the Animate 3D API, and receive animation assets in industry‑standard formats (FBX, BVH, glTF) ready for immediate import into Unity, Unreal Engine, or other pipelines. Real‑time body tracking and automatic retargeting to custom rigs accelerate iteration, while built‑in quality metrics ensure output consistency for production use. The platform’s subscription model provides scalable access to compute resources, API calls, and premium support, enabling creators of any size to integrate high‑quality motion data on demand.
Target Audience
The primary customers are indie game developers, XR artists, and small to mid‑size visual effects studios that need fast, affordable motion capture without dedicated hardware. Additionally, educational institutions and research labs use the platform for rapid prototyping of digital humans and biomechanics studies.
Features
- Deep convolutional pose estimation network that extracts 3‑D joint trajectories from 2‑D video with sub‑centimeter accuracy
- End‑to‑end facial capture pipeline using a separate CNN to generate blendshape coefficients and eye‑gaze vectors
- Cloud‑hosted processing cluster with GPU acceleration, delivering results in under 30 seconds for typical 10‑second clips
- Animate 3D RESTful API and SDKs for Unity/Unreal, supporting batch uploads, webhook callbacks, and on‑the‑fly retargeting to user‑defined skeletons
- Automatic rig retargeting engine that maps captured motion to industry‑standard rigs (Humanoid, Mixamo, custom) while preserving foot‑plant and inverse‑kinematics constraints
- Export options in FBX, BVH, glTF, and Alembic, with configurable frame rates and coordinate systems for seamless pipeline integration
- Built‑in motion quality scoring and anomaly detection to flag low‑confidence frames and suggest re‑capture
- Scalable subscription tiers with per‑minute compute quotas, enabling cost‑effective usage for indie creators up to enterprise studios