Kinetix

Kinetix is an AI research company that combines 3D techniques with machine learning to understand and generate human motion. Its research covers human pose estimation, computer vision and the foundational components of 3D character workflows, and it states it has spent roughly five years building foundational models for 3D human animation. The company describes its team as engineers, scientists and strategists.

The company's current flagship research output is Kamo-1, a 3D-conditioned video generation model intended to add physical grounding and controllability to AI video. Kamo-1 accepts multimodal conditioning signals including 3D camera trajectories, 3D character animations reconstructed from RGB acting videos, reference images and natural language prompts, and uses depth and pose estimates as control signals so that generated video is consistent with the underlying spatial layout and articulated motion. By jointly conditioning on camera trajectories and 3D animations it allows coordinated control of camera and character motion within a single shot; the pipeline reconstructs a scene, places a 3D character mesh, animates it over time, and renders the final video through a flow-matching generation process.

At its earlier stage, Kinetix was positioned as an AI-assisted 3D animation company offering a no-code platform that converted uploaded video clips into 3D animated avatars for use in virtual worlds and metaverse platforms, with a free consumer-facing product and an asset library that included content from Adobe Mixamo and Ready Player Me.

Business model

Kinetix licenses and integrates its models and core technologies into third-party platforms and products rather than only selling a standalone tool: its video-to-animation model is natively integrated into Unity's Muse AI services, packaged as an in-game "Emote Creator" feature in KRAFTON's OVERDARE (with an accompanying moderation system), and licensed into Adobe Mixamo. Earlier the company operated a free no-code web platform for creators converting video into 3D animated avatars.

Traction

Model integrations shipped with three named platforms — Unity Muse, KRAFTON's OVERDARE (Emote Creator plus moderation system) and Adobe Mixamo. In 2022 the company reported a team of 30 and partnerships with virtual worlds The Sandbox and ZEPETO, both of which also invested in its seed round.

Latest developments

The company's website presents Kamo-1, a 3D-conditioned video generation model supporting multimodal conditioning on camera trajectories, 3D character animations, reference images and text prompts, alongside deployed integrations in Unity Muse, KRAFTON's OVERDARE and Adobe Mixamo.