Mesh Motion Mapper

Blender existing modifier

Blender has tools to tranfer motion from one mesh to another, such as mesh deform modifier. The mesh motion mapper does not provide functions that does not exist in Blender. It is also not faster than Blender's internal modifier or geometry node given the addon system is going through Python. However, we create the mesh motion mapper node for a specific use case for transfering simulation result to a render mesh to simplify the setup process.

Motivation

The simulation mesh is almost never the same as the render mesh. In most of our examples, we used render mesh as subdivided simulation mesh. However, in production, many cases, we can not assume any topological relation between the render mesh and the simulation mesh.

Discrepancy between simulation mesh and render mesh

A reminder, HiPhyEngine can only guarantee intersection free of the simulation mesh. User will have to tune collision radius and barrier stiffness so there is enough space for the render mesh to be intersection free, when the motion is transferred.

To further complicate the issue, sometimes user may not wish to simulate the whole mesh. For instance, with the image below, we want to resolve the collision mesh's self intersection with motion mapper, however, we want to exclude hands, toes, and head from the simulation, because their small details will cause issues for the collision resolution algorithm, and they should be driven by animation directly anyway. Therefore, we created the simulation mesh by excluding those areas. However, the final render mesh will still need to have those parts of the body.

We created the mesh motion mapper to provide user a simple way to combine the motion from animation and simulation.

Shapes

Guide Shape

The Guide Shape is the main driver for the Driven Shape. Guide Shape does not have to have the same topology as the Driven Shape, however, it must be a triangle mesh. During binding, Driven Shape vertices will be bond to the surface to the Guide Shape if it is within the Binding Radius of a Guide Shape triangle face. Guide Shape is usually the simulation mesh.

Driven Shape

The Driven Shape is the mesh that you wish to deform and follow the Guide Shape. Driven Shape is bond to the Guide Shape per vertex, therefore the Driven Shape can be any type of mesh. If a vertex of the Driven Shape is not bond to the Guide Shape (due to out of binding radius), it will follow the Target Shape instead. Driven Shape is usually the render mesh.

Target Shape

The Target Shape is the mesh that Driven Shape will follow if a vertex is not bond to the Guide Shape (due to out of binding radius). Target Shape must have the same vertex count as the Driven Shape, however it can be any type of mesh as only the vertex information of the Target Shape is used. Target Shape is usually the raw animated mesh.

Properties

Property Name	Description
Binding Radius	Radius for binding guide and driven mesh (In guide mesh space)

Binding Space

The binding happens in the guide mesh space. So Binding Radius is also in the guide mesh space. If there is a scaling factor of 100 of the guide mesh, Binding Radius will be divided by 100. If you are uncertain how the conversion works, you can always just make sure the scaling of the guide mesh to be 1.

Example

You can find a mesh motion mapper example here.

In the scene: female_base is the soft collider simulation used for the guide shape, female_base.driven is the driven shape for rendering, and female_base.anim is the original animated mesh used as the target shape. Note that female_base has part of the mesh removed as they are not relevant to contact resolution. female_base.driven and female_base.anim shares the same topology.

That the Target Shape is the raw animation and contains self intersection at the upper arm.

The Guide Shape is the simulated soft collider. It resolves the self intersection but we have excluded the head from the simulation.

The Driven Shape is the render mesh. It combines Target Shape and Guide Shape, and it has the resolved arm intersection and animated head.