Globally turn all deformation on/off. This is useful when you want to adjust the capture parameters and want to see the original input geometry.

When this is on, the deformation only occurs within the capture region (this is the default). When this is off, it extends the deformation beyond the end of the capture region to the end of the model.

A global squash/stretch works best with this parameter off, since the squishing will be performed uniformly rather than tapering off to the capture bounds.

Apply bend symmetrically across the capture origin, in both the forward and backward directions.

Perform bend deformation, show the Bend handle in the viewport.

Whether the amount of bend is an angle, or a direction that the capture spine capture should end up pointing along.

When Mode is “Angle”, the angle in degrees to bend. You can make this over 180 to bend something over itself multiple times.

When Mode is “Direction”, bend the capture region spine to end up pointing in this direction.

Perform twist deformation, show the Twist handle in the viewport.

Angle in degrees to twist around the capture axis. This can be more than 360 degrees. The total twist is distributed evenly along the length of the capture region.

When Deform in Both Directions is enabled, will cause the continued deformation in the symmetric direction to be a continuation of the twist in the foreward direction.

Perform length scaling, show the Length handle in the viewport.

Do “squash and stretch”: as you scale the length, apply an inverse perpendicular scale, so lengthening the model makes it skinny, and shortening the model makes it bulge. (Because this is applied smoothly, it is not technically accurate volume preservation). When Limit deformation to capture region is on, the scaling falls off at the ends of the capture region.

Scale the length of the captured geometry by this factor.

Perform tapering, show the Taper handle in the viewport.

The axes to enable tapering along. These axes are in relation to the Up Vector (Y) and Capture Direction (Z).

Whether to blend linearly between taper scales, or smoothly (ease-in and ease-out).

The amount to scale the end of the capture region.

The amount to scale the “midpoint” of the capture region. You can move this scale between the start and end of the region using the Squish pivot parameter below.

The position of the “midpoint” scale along the length of the capture region, where 0 means the start of the region, 0.5 is halfway, and 1.0 is the end.

Lets you apply an arbitrary number of taper scales along the length of the capture region, controlled by a ramp.

When Enable ramp is on, the horizontal axis represents the length of the capture region (start on the left, end on the right). At each point across, the value represents the amount of scaling (where 0.5 means “no scaling”, values greater than 0.5 scale out, values less than 0.5 scale in). Note that you can set a value for a point above 1.0 using the parameters below the ramp.

The bend is always perpendicular to the capture axis, but this specifies what orientation it should use.

The up vector to use.

Angle (in degrees) to rotate the up vector around the capture spine.

Click this, then click two points in an orthographic viewport to define the capture region. This is the same as choosing the points when using the shelf tool.

The center of the source capture plane.

The normal of the capture planes, and the direction pointing to the destination capture plane from the source.

Controls the region of space that receives the partial transforms. Must not be zero in length.

When this is on, create this point float attribute on the output geometry. The attribute will have 0 on points that were not deformed, and 1 on points that were fully deformed. This may be useful for visualizing the amount of deformation in different parts of the model.

If this is on, and the second input is connected, use the point positions in the second input to compute which points are in the capture region. If this is on and the second input is not connected, use a rest point attribute if present (this should be a 3float attribute containing the “rest” position of each point). If this is on but no rest attribute exists, the node creates it from the initial point positions. If this is off, the node uses the point positions as they are in the input.

A space-separated list of names and/or patterns. Attributes whose names match are transformed along with the point positions. The node uses each attribute’s type info to transform it properly (as points, vectors, or normals).

If you change this from the default (*, all attributes), make sure P (point positions) is on the list unless for some reason you want to transform other attributes but not actually deform the geometry.

Recomputes normals of polygons that have both transformed and untransformed points.

If you're transforming the full geometry or all connected components, this setting is irrelevant. It only matters if you deform part but not all of a connected piece.

This does nothing if P is not listed/matched in Attributes to transform.

Normal lengths remain unaffected.