Video-to-animation AI
Kinetix has created state of the art Machine Learning models to make animation creation as easy as it should be.
Last updated
Was this helpful?
Kinetix has created state of the art Machine Learning models to make animation creation as easy as it should be.
Last updated
Was this helpful?
Kinetix AI allows for single camera motion capture. Our AI deep learning algorithms can extract any human motion from any single camera video.
Kinetix AI will analyze the video sent by the player, identify the human that's appearing, and read their movement, in 3D (recreating depth). Once the movement is fully analyzed, it will be recreated in 3D to generate an animation file. Then, the animation is retargeted on any custom avatar.
Kinetix tracks body movements at specific positions and then name them at each critical joints (e.g. feet, ankle, knee, shoulder, elbow). To get the best final output, Kinetix AI is making predictions accurate at the centimeter scale.
Hands movements are even more complex to detect but nothing’s impossible! At the moment, Kinetix AI is spotting if the fists are clenched, hands and fingers are opened or fingers are detached (celebrating victories ✌️for example). More complex hands movements are still hard to detect but Kinetix keeps improving its models at large scale.
Kinetix AI can also detect and isolate half body movements while keeping the other half static during the transposition process.
Which means you can create both lower body or upper body isolated animations while outputting a full body animation in the end.
Facial motion is currently not recognized by our algorithm.
Stay tuned! The feature is being developed and will arrive soon.
