A subset of points in the input geometry to run the program on. Leave this blank to affect all points in the input.

What the group is made of.

A snippet of VEX code that will manipulate the point attributes. You can use @‹variable_name› syntax to access geometry attributes.

Point and Vertex attributes that match this pattern will be transformed. Their Type Info will be used to determine how they should transform as vectors and normals need different operations. Attributes with Type Info of point or hpoint will not be transformed.

If P matches this string, then primitive transforms will also be rotated.

Recomputes any normals that are affected by polygons that use both transformed and untransformed points. When transforming the full geometry or full connected components, the results should be the same as when this option is off, i.e. the normals will just be transformed and not recomputed. This does nothing if P is not being transformed.

Normal lengths remain unaffected.

To transform attributes other than position, such as velocity or normals, how the deformation changes over space must be known, not just how the deformation treats a single point. If you know how to compute this, you can write the derivative into the xform parameter and select Explicitly Set Xform. Usually, however, deformations can have difficult to compute behavior, so it is best to let this node automatically compute the deformation behavior.

Four Points does, as its names suggests, three extra deformations to determine how the deformation changes in x, y, and z directions. This is faster than Central Differences, but is biased as it only looks in one direction.

Central Differences does six extra deformations to find a balanced change in a neighborhood around the point being deformed. This gives the best results for the automatic computation, but must invoke the deformer twice as often.

A small step is taking in each direction to determine how the deformer is twisting space. This must be small enough to pick up the key feature changes of your space, but not so small as to lead to numerical problems. A good rule of thumb is half the inter-point spacing of your model.

The automatically computed or manually computed matrix may have scale or shear components in it. This will project them out, forcing it to be a rotation matrix. polardecomp is used for this purpose.

The sign of the derivative matrix can flip in certain regions of the transform. This will detect that and flip the sign back so the determinant of the matrix is always positive.

Only create attributes if their names match this pattern.

The default is to not allow creation of any attributes.

Bound attributes, such as vtxnum, cannot be created and will be ignored.

Requires that you declare @ bindings in snippets as prototypes before using them. This applies to both attributes (for example @Cd) and “convenience” bindings such as @ptnum and @Frame. For example:

// Declare bindings
int @ptnum;
float @Frame;
vector @Cd;

// Use bindings after declaration
int pointnum = @ptnum;
float red = @Cd[0] / @Frame;

Automatic binding with the @ syntax can be convenient, but as your scene becomes more complex there is the risk that a typo in an @ binding will silently just bind a non-existent attribute.

VEX functions like ch() usually evaluate with respect to this node. Enter a node path here to override where the path search starts from. This is useful for embedding in a digital asset, where you want searches to start from the asset root.

When a VEX parameter is exported, the bound attribute will be created if it doesn’t exist. This pattern can be used to override the export option on the VEX shader to avoid writing to or creating certain attributes. The pattern matches the VEX parameter, not the bound attribute. The attribute will still be bound for reading.