As opposed to the Point op or other ops that manipulate control points (CVs), the Clay op operates on the primitive contours themselves, providing a direct, intuitive, and unconstrained way of reshaping geometry. Thus, rather than translating CVs to change the aspect of the primitive, the Clay op takes the inverse approach of repositioning the CVs to reflect the tug on the primitive’s skin.

The point that defines the area to be modified is called a “target point” or “target” for short. It is expressed as a (u,v) pair in the parametric space of the primitive and ranges between 0 and 1 in both U and V. The image of the target point on the primitive is a 3D point which Clay can displace in several ways. Furthermore, if the primitive is a surface, there is are options to pull only the point or a whole isoparametric curve in either U or V.

Clay does not refine the faces and surfaces unless asked to, so the complexity of the geometry does not increase. The area affected by the change varies with each primitive type and topology. In all cases it is possible to reduce the amount of change by inserting a Refine op before the Clay op and inserting detail around the target point. For other ways to increase the locality of the deformation as well as its sharpness, see U and V Sharpness below.

If a second input is present, it is possible to snap the target (u,v) point to an (s,t) point on the first primitive of the second input. Without a second input, the primitives can be made to snap to themselves. Moreover, the Clay op is able to snap the target to arbitrary points in space.

Both this op and the Align SOP can be used effectively as snapping tools and building blocks for curve networks. The main difference between the two ops is that Clay deforms the inputs partially, while Align translates and/or rotates the whole primitive.

The Clay op accepts a mix of any combination of face and surface types.