The MetaHuman Performance asset editor is where you configure what to solve, run the processing, and export your animation as capture data. You can solve for face, body, or both in a single pass from the same footage.
Create and Configure the MetaHuman Performance Asset
To turn your capture data into animation for your digital-human you need a MetaHuman Performance asset.
To get started:
In the Content Browser, right-click and select MetaHuman > MetaHuman Performance.
Double-click to open the asset, and click the save icon.
In the Details panel, select your ingested footage in the Footage Capture Data field.
Choose the type of performance tracking in the Processing Parameters section of the panel.
Set What Performance to Solve
By default, face animation is enabled Processing Parameters section. Adjust the following settings based on what you want to capture.
Face Animation Only
Leave the default settings unchanged. You can toggle settings as needed, such as Solve Tongue and Head Stabilization.
In the Advanced settings, there is an option to select the Facial Models used to process the Monocular Pipeline. The best models for each use case are selected automatically. You can manually override the automatically-selected model.
The table lists the available models.
| Model | Description |
|---|---|
Face Detector | Processes facial footage only. The Face Detector model is selected. However, if the MetaHuman Animator Body Animation plugin is enabled, an additional Small Face Detector model is available. In this case, if you enable Body Tracking, the Small Face Detector model is used by default. This face detector works better when the face occupies a small fraction of the image. If the Body Tracking is disabled, the standard Face Detector model is selected by default. |
Face Head Pose Tracker | A single Head Pose Tracker model is available. |
Face Solver | Processes of facial footage only, the default model selected depends upon the Device Class of the Footage Capture Data. If the class is Stereo HMC, the Stereo HMC Rig Solver model is selected, which works better on (mono) footage from a greyscale Visible Light or IR Stereo HMC capture device. For all other cases, the original Generic Rig Solver model isselected. However, if the MetaHuman Animator Body Animation plugin is enabled, an additional Small Face Solver model is available. In this case, if you enable Body Tracking, the Small Face Solver model is used by default. This face solver works better when the face occupies a small fraction of the image. If Body Tracking is disabled, either the Generic Rig Solver or Stereo HMC Rig Solver is selected by default using the rules for facial footage above. |
Body Animation Only
Uncheck Facial Tracking, then check Body Tracking. Set Head Movement Mode to Disable.
Face and Body Animation Together
Enable Facial Tracking and Body Tracking. Then go to Processing Parameters > Facial Tracking > Advanced > Models and make the following changes:
Set Face Detector to Small Face Detector.
Set Face Solver to Small Face Solver.
If Small Face Detector or Small Face Solver don't appear in the dropdown, enable Show Engine Content in the Content Browser filter settings.
Process and Export
To process and export the animation:
(Optional) Set the Visualization Object to your MetaHuman. Use this to preview the solve result before exporting.
The asset type changes based on the processing type you choose.
Adjust the timeline to trim the portion of the footage you want to capture.
In the toolbar, click Process and wait for the solve to complete.
Confirm the sequence looks correct. You can adjust the timeline as needed and click Process again.
In the toolbar, click Export Animation.
Enter a name and location for your animation sequence, then choose how to export the skeleton.
Existing Skeleton exports the animation onto a MetaHuman skeleton, with retargeting applied automatically. Select the skeletal mesh of your assembled MetaHuman in the Target Skeleton or Skeletal Mesh field. Use this when your target character is a MetaHuman.
Performer Skeleton creates a new skeletal mesh based on the actor's proportions from the footage. Use this when you plan to retarget the animation to a non-MetaHuman character.
