The orientation of each point in /obj/geo1/dopnet1 can be read from the “orient” point attribute. The attribute value is a quaternion describing how the geometry has rotated relative to its rest pose. /obj/geo1/copytopoints2 uses this attribute to orient each control.
I think I now understand what you are trying to do. Since the original geometry does not contain an “orient” attribute, the Wire Solver assigns a default value (the identity rotation) which isn't aligned to your geometry on most frames. You should provide an explicit “orient” attribute (see /obj/example/attribvop1 in the attached example).
Is it possible to find the position of a point in the world ?
Because for cable or other grain simulation, it's complicated for me to recover an animation of the object. I come from 3ds Max where a simple link does the job.
If I understand what you are trying to do, the initial orient values do not match what you want them to represent. The Copy SOP applies the minimal rotation to align the geometry's Z-axis with each template point's normal but you want the orient values to reflect the rotation described by /obj/coil/transform2.
Disconnect the transform2 node then place it between the copy1 and OUT_wire nodes. This will cause the rotation to be applied to the orient values as desired.
I am sorry, I thought I had found an example that would explain a bug on my real scene. But no.^^
In reality, the lines are animated and then imported into Houdini in alembic. I am then obliged to extract the orientation of the intrinsic ones for the copy. But the simulation becomes wrong. May be because of the difference in orientation between these two software.
In the meantime I found a solution: I merge the orient after the simulation with qmultiply().