NEXT-SYSTEM implemented their independently developed AI pose estimation engine "VisionPose" into their App "MICHICON-Plus", to demonstrate how the AI engine can be utilized in smartphone and tablet applications.
"MICHICON-Plus" is a full-body real-time motion capture application that captures and reflects motion data on a 3DCG character, and supports motion data export. It provides an easy way for motion capture without using any equipment apart from a smartphone/tablet camera, and can be used for VTuber video creation and similar.
What is "MICHICON-Plus" and How Does it Work?
"MICHICON-Plus" is a motion capture application that uses the latest technology to capture movements of the whole body of a person with the help of AI pose estimation engine "VisionPose", and projects these movements on a 3DCG character.
When performing motion capture, it is usually necessary to prepare multiple cameras and attach devices such as markers to the target. But with this app, no special equipment or professional cameras are needed, since the AI engine VisionPose can detect the user's skeleton information via only smartphone or tablet camera.
It also supports exporting motion data in CSV format, and changing the avatar to other characters through the online platform VRoid Hub*. In addition to real-time motion capture, the app also provides the option of projecting movement from videos on the user's device onto the 3DCG character. Additionally, the AR mode allows the user to make the characters appear as if they are in the real world by using camera images as the background.
For adding characters and exporting motion data in CSV you need to upgrade to the PRO version for a monthly fee (non-renewing subscription, $2.99 per month).
Here is a video showing what the app looks like:
Comment from our Engineers
We hope that "MICHICON" will provide an opportunity for users to get to know our AI pose estimation engine VisionPose. We will continue to develop the app to make it more user-friendly, including the UI.
Specifications and System Requirements
- Application Name: MICHICON-Plus
- Category: Entertainment
- Supported OS: iOS 13.0 and newer models
- Price: Free
- AppStore: MICHICON-PLUS on the App Store (apple.com)
Recommended Devices
- iPhone XS/XS MAX/XR
- iPhone 11～
- iPhone SE (2020)
- iPad mini (5th〜) *AR function is not supported
- iPad Air (3rd～) *AR function is not supported
- iPad Pro 11inch (1st～)
- iPad Pro 12.9inch (3rd～)
- iPad (8th〜)
Devices with A12 or later processors are recommended.
Please note that operation on devices not listed above is not covered by warranty or support.
Basic Functions
- Motion Capture Analysis
- AR Mode
- Green Screen Option
- Projecting movements from videos onto characters (max. 15 seconds)
- Switching Camera (front/rear)
- CSV Output (PRO version only*)
- Changing Characters (PRO version only*)
PRO version: non-renewing subscription, $2.99 per month
What is VisionPose?
VisionPose is an AI pose estimation engine that detects human skeleton information on real-time camera images, still images and videos, without the need for markers or special equipment. The AI is capable of detecting skeletal information of 30 key points in 2D and 3D in real-time and comes with two ready-to-use applications in addition to the SDK, so the user can start analyzing immediately after setting up their PC. As an all-around SDK package, it has already been applied to various fields in development and research, and has been implemented by over 250 clients inside of Japan. Sale outside of Japan has started with two products, VisionPose Standard for Windows C# and C++, and the number of supported platforms will be expanded sequentially for Linux/iOS/Jetson as well.