Hi
I am looking for a way to let a group of points further behave as a child of another point, after this masterpoint has been altered (say through noise or other secondary effects).
The transformeffect on this masterpoint should be re-applied to the other point group, with the masterpoint itself as a pivotpoint
Something gives me a hunch that inverse matrix transform might do it but I am not getting very far with it.

I hope someone could point me in the proper direction.
I have a file attached which should abstract the idea

Thanks for the thinks

Michiel

Is your ‘pc’ node an attempt of what you couldn't do with your ‘null1’ and ‘box1’ nodes?

Because if I understand correctly what your trying to do, all you need to do on ‘box1’ is to copy a relative reference from the null1 nodes Translate parameters to the Box1 Pivot Translate parameters.

Same with Rotate and Pivot Rotate.

Like what I've done with your scene (reduced by eliminating the pc node).

Hi Babaj

What is in the ‘pc’-node indeed represents the situation I am actually in. That being I don't have access nor control anymore over the incoming point-data. In this example-scene that access of course still exists but that was only for visualization. I do want to alter though parts of this data and have certain parts of the pc follow this alteration as if they'd still be connected.
Therefore I am looking for a way to calculate the relative altered offset of the master-point to its unaltered state and re-apply it to the others.

In your pc node you have the netbox colored red in which you say that is what you want but not through those means.

What is the limitation of that particular set-up? That you feel you don't have control over incoming ‘point-data’?

I don't understand why in the red netbox you are importing the ‘master point’ from the null then applying noise, and then copying the box points with the copy to points.

Why not just apply that noise to the ‘master point’ from the start. Like I've changed in your file here.

But as far as your original question is concerned. Things like vector subtraction can give differences in relative position.

The Pivot Transform example I gave can be applied in the same way at the sop level with the transform node set to groups to isolate only those you want it applied too.

The same can be done in vex with the maketransform function.

I'm not sure why your starting with a Transform at the object level (your object level null).

The only time I see that done usually is people doing rigging.

Hi Babaj

Your guess is spot-on, rigging at sop-level is what I am after.
Thanks for the heads up. Indeed, if the control over incoming transformation is mine, these things can be set-up much more accessible. However, my system should also work when I only have a disk-written pointcloud with attributes available, where the original underlying hierarchy is hidden to me. Therefore I made the example this way. In that scenario, to keep the (for example) fingers consistent when the wrist gets post-modified, I need transform comparisons, I guess.

Cheers
Michiel