orienting copies / polyframe question

[rbesca]

hey everyone! I'm trying to copy and orient those parts correctly: 

I used polyframe and created tangent normals and an up vector which works great for copying all the parts in place with the desired orientation.

but the top part just rotates with the whole thing when translating the circles they are copied to.

Can I use orient to lock them always pointing up? the tubes are supposed to connect to a disc above

polyframe for tangent N gives almost desired result

orientation_copy.hiplc

 

Edited January 26, 2020 by rbesca

[Noobini]

like this ?

it should simply be aligned to Z first, similar to what you've done to the longer tubes

then it would be correctly copied to points with N...not up...ie. I added a wrangle to make N = your up.

vu_orientation_copy.hiplc

[rbesca]

yup i tried this too, not that exact same way but I got this result when I connected the earlier output without polyframe, just the points I want to copy it to.

 

but now the parts don't align anymore at the bottom. if I try to offset it with pre-transform it will work for one of them and the rest is messed up

 

other method gives aligned result but messed up rotation

 

Edited January 26, 2020 by rbesca

[rbesca]

 

[Aizatulin]

Hi,

or you set the direction separately. For the smaller types you can use v@N = {0,1,0} and as up-vector cross(v@N, D), where D is the direction to neighbour point of the diagonal line. For the bigger type v@N = D and up-vector the same (or +- of it).

orientation_copyX.hipnc

[rbesca]

I don't understand everything quite yet but this is amazing, thanks! why does the beta rotation in the rotation matrix channel need a different factor? that channel uses the up vector instead of N so I assume it has something to do with that? I changed it around but things still worked, the only thing that changed that I could tell was the amount of full rotations it does when changing the slider?

[Aizatulin]

The matrices where more for me to check if the parameters worked correctly. You can ignore them.

To explain how they work:

• the first rotation is around the main axis (the direction, which is N)
• the second rotation is around the secondary axis (the up vector)
• the third rotation is around the main axis again (but it should not has transformed yet)

Each step is applying a rotation matrix on the base rotation R, which start with the identy matrix.

here is another example to visualize the rotation, where N = {0,1,0} and up = {1,0,0}. As you can see, that the rotation of the first matrices are now performed in the transformed space.

 

mat_chain_rotation.hipnc

Edited January 28, 2020 by Aizatulin

[rbesca]

ooh okay! thanks a lot for the explanation, this will get me there, powerful stuff