Using Pre-Viz
PREREQUISITE:
Pre-vis cannot be activated until your project has a successful intrinsic and extrinsic calibration. If you haven't completed this step yet, refer to the Capturing Intrinsic and Extrinsic Calibration documentation before proceeding.
- Starting Motion Capture
- Open an existing project, or continue in the project where you completed your intrinsic and extrinsic calibrations.
- Click ‘Pre-Viz’ to open the settings.
- Note: Projects are stored in the Home/moveai/projects_live file directory.
- Set the Number of Actors
- Select how many actors you want to track (1–4).
- Note: Avoid changing this setting while Pre-vis mode is active.
- Choose a Tracking Area
- This defines the detection zone, who the system is able to see and detect.
Option Description Camera Positions Creates a detection area based on the perimeter formed by the camera positions Polygon Lets you define a custom-shaped detection area by setting the number of sides and radius None Anyone visible to any camera will be detected
- This defines the detection zone, who the system is able to see and detect.
- Choose an Initialisation Mode
- This determines which detected actors the system will actively track.
Option Description Auto Automatically begins tracking any detected actor Click The operator manually clicks an actor's bounding box to begin tracking Hands Tracks actors who raise both hands above their shoulders
- This determines which detected actors the system will actively track.
- Activate
- Click Actviate
- Step into the capture volume inside Genesis, indicated by the green outline.
- The 2D detection points will lock onto the actors in the space, and actor tracks will begin loading. Once the actor track is created, the mesh will appear.
- Note: The mesh will be hidden if your 3D overlay is turned off.
Optimising Settings
The Settings panel is divided into three sections: Mocap Settings, Camera Settings, and General Settings.
Filtering
You can adjust the level of filtering to bias for either lower latency or higher quality mocap. Head to File > Settings to adjust the filtering strength from 0 (least filtering) to 5 (maximum filtering).
Note: A higher filtering value will produce smoother tracking but will increase latency. A lower value will reduce latency but may introduce more noise in the data.
Filtering Settings
- Smoother (Body)
- Sets the smoothing algorithm applied to body tracking data.
- Filtering Strength (Body)
- Adjusts the intensity of the body smoothing filter.
- Smoother (Hands)
- Sets the smoothing algorithm applied to hand tracking data. Default: Polynomial.
- Filtering Strength (Hands)
- Adjusts the intensity of the hand smoothing filter.
- Finger Tracking
- Enables or disables finger tracking.
Capture Mode
- Quality (Full body + fingers)
- This mode features a 65-keypoint model with finger tracking and is recommended for high-spec computers that meet our advanced system requirements.
- Performance (No fingers)
- This mode features a 29-keypoint model with finger tracking and is recommended for lower-spec computers that meet our standard system requirements.
Skeleton Estimation
| Setting | Description |
|---|---|
| Estimation Frames | The number of frames used to estimate the actor's skeleton |
| Sampling Interval | The interval between sampled frames during estimation |
When selecting a skeleton estimation mode, consider whether fast initialization is a priority. While longer estimation times take more time to start tracking, they deliver higher accuracy for actors entering the space.
Track Auto Removal
| Setting | Description |
|---|---|
| After losing sight | Removes the actor track when the actor is no longer visible to the cameras |
| After leaving tracking area | Removes the actor track when the actor exits the defined tracking area |
| Timeout (seconds): Once the removal criteria is met | The delay before an actor track is automatically removed. |
.png?width=620&height=633&name=image%20(13).png)
General Settings
| Setting | Description | Default |
|---|---|---|
| Render FPS | Sets the frame rate at which the viewport renders | 30 |