Make BezierSegment support N-dimensional beziers #9686
Conversation
We can simply take the dimension from that of the first control point. There is no functional change for existing callers. Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
| @@ -163,31 +163,29 @@ class BezierSegment(object): | |||
| def __init__(self, control_points): | |||
| """ | |||
| *control_points* : location of contol points. It needs have a | |||
| shpae of n * 2, where n is the order of the bezier line. 1<= | |||
| n <= 3 is supported. | |||
| shpae of n * D, where n is the order of the bezier line | |||
There was a problem hiding this comment.
"It needs to have a shape"?
| self._orders = np.arange(_o) | ||
| _coeff = BezierSegment._binom_coeff[_o - 1] | ||
|
|
||
| _control_points = np.asarray(control_points) | ||
| xx = _control_points[:, 0] | ||
| yy = _control_points[:, 1] | ||
| xyz = [_control_points[:, i] for i in range(0, dim)] |
There was a problem hiding this comment.
If this is an array, then isn't this just _contrrol_points.T?
|
|
||
| self._px = xx * _coeff | ||
| self._py = yy * _coeff | ||
| self._pxyz = [xx * _coeff for xx in xyz] |
| _y = sum(tt * self._py) | ||
|
|
||
| return _x, _y | ||
| _xyz = [sum(tt * px) for px in self._pxyz] |
There was a problem hiding this comment.
and this is np.sum(tt * px, axis=1).
|
I'm happy to update this as you suggest. Also, I'd be happy to write a test but the test suite does not run for me. See error message below. I would try running just the existing test to see if I can extend it but AFAICT there is no existing unit test for this class. |
|
Using I'm not sure if there's an existing test; it might just be done implicitly through other usage. |
|
Closing as superseded by #13355. Thanks for the PR! |
Here is a simple patch that generalises matplotlib.bezier.BezierSegment to N-dimensional segments. I needed this for my own activities, not for anything in matplotlib, but the matplotlib bezier was convenient.
I have forward ported this patch from Debian's matplotlib 2.0.0+dfsg1-2. I have tested this change only in the context of 2.0.0, and only ad-hoc. But it's very straightforward so I'm hoping that the CI will be happy.