HeadStabilization
Only in Wrap4D
Finds stabilization transform for input expression using forehead and nosebridge masks and a neutral geometry and optional by screen points.
Attention
The node is not guaranteed to give good results if Expression and Neutral geometries are not aligned, or their scale is different from our gallery.
Inputs
- Expression Geometry
Geometry
Geometry to be stabilized- Neutral Geometry
Geometry
Geometry to serve as a stabilization reference- Forehead Mask
VertexMask
Mask that covers the forehead area- Nosebridge Mask
VertexMask
Mask that covers the nosebridge area- Reference Geometry
Geometry
Geometry to serve as a stabilization reference- Reference Points
NamedPointsOnTriangle
Points on reference geometry that will be compared with screen points- Target Cameras
Camera
Cameras for screen points- Target Screen Points
ScreenPoints
Points on images that will be used for stabilization- Target Images
Image
Images of the target from cameras
Output
Geometry
Expression geometry with stabilization transform on it
Parameters
- Max Iterations:
Max number of stabilization iterations
- Trust Region:
Number representing how much we trust the difference between Expression and Neutral geometries. Higher values may help if the result appears unstabilized. Values lower than the default are generally unadvisable
- Forehead Coef:
Number representing the weight of actor’s facial features, specifically forehead. The default value usually provides a good result, but might be increased or reduced depending on the actor
- Nosebridge Coef:
Number representing the weight of actor’s facial features, specifically nosebridge. The default value usually provides a good result, but might be increased or reduced depending on the actor
- Smoothing Iterations:
Number of smoothing iterations to remove noise from the geometry
- Order:
Number representing the amount of averaging that happens during stabilization. The default value usually provides a good result, but higher values may result in better stabilization of expressions with strong nose compression. Higher values also make the algorithm slower
- Control Points Weight:
Fitting weight, corresponding to connecting points to screen points
Vertex Mask Examples
- Forehead Mask
- Nosebridge Mask