animatrix

Attribute Interpolate and Transform by Attribute SOPs

Added 2020-08-10 07:25:42 +0000 UTC

These are two of my favourite utility SOPs. They can make the handling of multiple attributes of all types and classes much easier, thus avoiding to write convoluted VEX code to do the same.

As you know VEX doesn't have generics like C++ or C#. Therefore handling multiple attributes of different types cause unnecessarily long and complex code. When we have to handle multiple classes, typically this results in multiple Attribute Wrangles or Attribute VOP SOPs.

Particularly Transform by Attribute SOP with matrices can be really powerful, which is why I also updated my Rigid Transform SOP to use this operator so that it can modify all appropriate attributes for free, not just P.

I would recommend utilizing these operators when you can when designing new HDAs, rather than trying to perform the same operations manually in VEX, for clarity, brevity and max performance.

That's also why I will update my deformer VOPs to output matrices that can be used in the same manner, making them even more generic, and versatile.

In an ideal world, it would be even better to implement them as space warps, but unfortunately it requires a new primitive type called analytic fields/volumes, which Houdini doesn't have.

The idea is that you define a field given some dimensions, a location, and an implicit function derived from the parameters, such that anywhere you sample this field, the implicit function is evaluated at that position to return a value. For example an analytic falloff field. It's not rasterized like a fog volume that we have to use now. Therefore it's uses very little memory, it's more precise and resolution independent.

But until then, this will do.