3. Custom template
Helenhayes
First, follow the Helen Hayes model's marker placement method in real-time mode and record a set of unnamed marker data. During data recording, ensure that all markers are visible and unobstructed. You may perform simple actions such as twisting the waist, waving a hand, lightly lifting a leg, side-kicking, back-kicking, rotating the wrist, turning the ankle, nodding, tilting the head left and right, and squatting.
After recording, load this dataset in post-processing mode and ensure the data is at frame one.
Next, add the Helen Hayes template provided by us to the assets panel and use "Quick Identification" to identify the unnamed markers in the first frame of the data. Ensure all markers in the first frame are correctly recognized.
Pause the data at the first frame and highlight all frames on the data axis. Click the "Correct" button to identify and correct markers in all frames. After correction, markers in all frames should be recognized. Drag the data axis to check whether the markers in all frames have been correctly identified.
If markers are not correctly identified in a certain frame, use "Quick Identification" to recognize the unnamed markers. If the same marker is not identified across consecutive frames, select the range and use "Quick Identification" on the first frame within the range. After identification, click the "Correct" function to recognize the unnamed points in the selected range.
After all markers are identified in all frames, go to the first frame and click the "Solve" button. After the solving process, the links between points, the human skeleton, and the "V Markers" in the template will be displayed.
If there are still anomalies in the data, refer to the "Data Processing" section, subsections four, five, six, and seven, for instructions on using the data repair functions to fix the data.
Click "File" and then "Save As Model." In real-time mode, load the saved model, have the model walk into the motion capture area, and the Helen Hayes model will display in 3D.
Baseline + Toe, Headband (53)
Record unnamed marker data for the 53-marker full-body model in real-time mode. During data recording, ensure all markers are visible and unobstructed. Perform simple actions such as twisting the waist, waving hands, lightly lifting a leg, side-kicking, back-kicking, rotating wrists, turning ankles, nodding, tilting the head left and right, and squatting.
Load this set of recorded unnamed marker data in post-processing mode, and load the human body template provided by us in the assets panel.
Follow the same steps as mentioned above for customizing the "Helen Hayes" human body template.
Creating Custom Templates
Custom templates can be created by applying markers to objects such as robotic arms, robotic dogs, jump ropes, badminton rackets, etc. After creating a custom template, post-process the data and then load the custom model file in real-time mode for data collection and analysis.
Apply markers to the object to be captured. Maintain a reasonable distance between markers and avoid regular patterns to prevent misidentification.
Move the marked object into the motion capture area and record its movement sequence. Ensure that markers are not lost or obstructed during recording. Then, load the recorded data in post-processing mode.
Open the assets panel, click the "Add Asset" button to create a new asset, enter the asset name, select "Human, " and then click "Create Object."\
Name markers in the "Markers" tab of the assets panel. Go to the first frame of the data, select all frames, and open the "Quick Identification" dialog to identify unnamed points on the object. After identification, create links between points.
Decide whether to create a skeleton based on your needs. For instructions on creating a skeleton, refer to "Twelve, Data Processing (II): Establishment of Markerset (Rigid)."
Go to the first frame of the data, select all frames, and click "Track Identification." All points on the object will be identified in subsequent frames. If a skeleton has been created, click "Solve, " and the skeleton will be displayed in all subsequent frames.
If some frames have unrecognized points during playback, refer to sections four, five, and six in "Twelve, Data Processing" for instructions on repairing the data.
After repairing the data, save the model and the motion capture data. Switch to real-time mode, load the repaired model file in the assets panel, connect the cameras, and play the footage. When you place the object in the capture area, the 3D view will display the model.
If model recognition is unstable in real-time mode, it may be due to a sudden increase in the distance between named points on the object during motion. If this occurs, collect another set of data where named points change abruptly or exhibit anomalies, and remedy the data again in post-processing mode.
