Rows to columns

[Andrew Kant]

I'm still a novice in Houdini. Help me understand how to swap the rows and columns of primitive indices. It is necessary that I can rotate the indexes, as shown in blue.

[Atom]

Try dropping down a Sort node and place it in Shift mode. Then you can offset primitives with the slider.

Edited September 4, 2017 by Atom

[Andrew Kant]

1 hour ago, Atom said:

Try dropping down a Sort node and place it in Shift mode. Then you can offset primitives with the slider.

Sort Node gives the wrong result. I do not need to shift the indexes, but rotate them 90 degrees. Node Sort works well with Grid, you can specify which axis to build columns, but for the torus it does not fit. I think that this can be implemented in the VOP. But I do not know.

Edited September 4, 2017 by Andrew Kant

[f1480187]

Compute index attribute:

// Primitive wrangle.
int rows = chi("../torus1/rows");
int cols = chi("../torus1/cols");
i@index = @primnum % cols * rows + @primnum / cols;

Then Sort by attribute:

swap_rows_cols.hipnc

 

By the way, this is reshape and transposition. Not 90 degree rotation:

| 0  1  2  3  4  5|
| 6  7  8  9 10 11|
|12 13 14 15 16 17|

| 0  3  6  9 12 15|
| 1  4  7 10 13 16|
| 2  5  8 11 14 17|

 

Edited September 4, 2017 by f1480187

[Andrew Kant]

1 hour ago, f1480187 said:

Compute index attribute:

// Primitive wrangle.
int rows = chi("../torus1/rows");
int cols = chi("../torus1/cols");
i@index = @primnum % cols * rows + @primnum / cols;

Then Sort by attribute:

swap_rows_cols.hipnc

 

By the way, this is reshape and transposition. Not 90 degree rotation:

| 0  1  2  3  4  5|
| 6  7  8  9 10 11|
|12 13 14 15 16 17|

| 0  3  6  9 12 15|
| 1  4  7 10 13 16|
| 2  5  8 11 14 17|

 

Many thanks! That's exactly what I wanted. I was also advised to use the Transpose node. Can you know how you can apply it for this task?

[f1480187]

It's limited to Houdini's native 2×2, 3×3 and 4×4 matrices, and it probably used for shading and 3D transformation tasks. We can't really use it to transpose r×c matrix. Simple numerical algorithm in VEX did the job too.

 

For the sake of completeness, to work with arbitrary-sized matrices in Houdini use numpy:

import numpy as np

rows = hou.evalParm('../torus1/rows')
cols = hou.evalParm('../torus1/cols')

# Make initial matrix representation.
old_sort = np.arange(rows * cols).reshape(rows, cols)

# Reshape and transpose.
new_sort = old_sort.reshape(cols, rows).T

# Make index attribute.
node = hou.pwd()
geo = node.geometry()
geo.addAttrib(hou.attribType.Prim, 'index', -1)
geo.setPrimIntAttribValues('index', new_sort.ravel())

 

swap_rows_cols_python.hipnc