nibabel_apply_affine.Rmd

---
jupyter:
  jupytext:
    notebook_metadata_filter: all,-language_info
    split_at_heading: true
    text_representation:
      extension: .Rmd
      format_name: rmarkdown
      format_version: '1.2'
      jupytext_version: 1.13.7
  kernelspec:
    display_name: Python 3
    language: python
    name: python3
---

# Applying coordinate transforms with `nibabel.affines.apply_affine`

We often want to apply an affine to an array of coordinates, where the last
axis of the array is length 3, containing the x, y and z coordinates.

Nibabel uses `nibabel.affines.apply_affine` for this.

For background see: The nibabel.affines module.

```{python}
import numpy as np
from nibabel.affines import from_matvec, to_matvec, apply_affine
```

```{python}
points = np.array([[0, 1, 2], [2, 2, 4], [3, -2, 1], [5, 3, 1]])
points
```

```{python}
zooms_plus_translations = from_matvec(np.diag([3, 4, 5]),
                                      [11, 12, 13])
zooms_plus_translations
```

```{python}
apply_affine(zooms_plus_translations, points)
```

Of course, this is the same as:

```{python}
mat, vec = to_matvec(zooms_plus_translations)
(mat @ points.T).T + np.reshape(vec, (1, 3))
```

The advantage of `nib.affines.apply_affine` is that it can deal with arrays
of more than two dimensions, and it transposes the transformation matrices for
you to apply the transforms correctly.

A typical use is when applying extra affine transformations to a X by Y by Z
by 3 array of coordinates.