Skip to content

armgit / manim

Public Imported from GitHub
0

A community-maintained Python framework for creating mathematical animations.

Fix performance issues with :meth:`~.VMobject.get_arc_length`, stemming from :pr:`1302` (#1315)

Anton Ballmaier966572140b2c
diff --git a/manim/mobject/types/opengl_vectorized_mobject.py b/manim/mobject/types/opengl_vectorized_mobject.py
index 171ae46af..90cf465ee 100644
--- a/manim/mobject/types/opengl_vectorized_mobject.py
+++ b/manim/mobject/types/opengl_vectorized_mobject.py
@@ -1,6 +1,7 @@
import itertools as it
import operator as op
from functools import reduce, wraps
+from typing import Optional
import moderngl
import numpy as np
@@ -551,7 +552,7 @@ class OpenGLVMobject(OpenGLMobject):
return bezier(self.get_nth_curve_points(n))
def get_nth_curve_function_with_length(
- self, n: int, n_sample_points: int = 10
+ self, n: int, sample_points: Optional[int] = None
) -> typing.Tuple[typing.Callable[[float], np.ndarray], float]:
"""Returns the expression of the nth curve along with its (approximate) length.
@@ -559,7 +560,7 @@ class OpenGLVMobject(OpenGLMobject):
----------
n
The index of the desired curve.
- n_sample_points
+ sample_points
The number of points to sample to find the length.
Returns
@@ -570,9 +571,12 @@ class OpenGLVMobject(OpenGLMobject):
The length of the nth curve.
"""
+ if sample_points is None:
+ sample_points = 10
+
curve = self.get_nth_curve_function(n)
- points = np.array([curve(a) for a in np.linspace(0, 1, n_sample_points)])
+ points = np.array([curve(a) for a in np.linspace(0, 1, sample_points)])
diffs = points[1:] - points[:-1]
norms = np.apply_along_axis(get_norm, 1, diffs)
@@ -600,10 +604,15 @@ class OpenGLVMobject(OpenGLMobject):
yield self.get_nth_curve_function(n)
def get_curve_functions_with_lengths(
- self,
+ self, **kwargs
) -> typing.Iterable[typing.Tuple[typing.Callable[[float], np.ndarray], float]]:
"""Gets the functions and lengths of the curves for the mobject.
+ Parameters
+ ----------
+ **kwargs
+ The keyword arguments passed to :meth:`get_nth_curve_function_with_length`
+
Returns
-------
typing.Iterable[typing.Tuple[typing.Callable[[float], np.ndarray], float]]
@@ -613,16 +622,18 @@ class OpenGLVMobject(OpenGLMobject):
num_curves = self.get_num_curves()
for n in range(num_curves):
- yield self.get_nth_curve_function_with_length(n)
+ yield self.get_nth_curve_function_with_length(n, **kwargs)
def point_from_proportion(self, alpha):
- curves_with_lengths = list(self.get_curve_functions_with_lengths())
+ if alpha == 1:
+ return self.get_points()[-1]
+
+ curves_and_lengths = tuple(self.get_curve_functions_with_lengths())
- total_length = np.sum(length for _, length in curves_with_lengths)
- target_length = alpha * total_length
+ target_length = alpha * np.sum(length for _, length in curves_and_lengths)
current_length = 0
- for curve, length in curves_with_lengths:
+ for curve, length in curves_and_lengths:
if current_length + length >= target_length:
if length != 0:
residue = (target_length - current_length) / length
@@ -678,15 +689,26 @@ class OpenGLVMobject(OpenGLMobject):
)
return points[distinct_curves.repeat(nppc)]
- def get_arc_length(self, n_sample_points=None):
- if n_sample_points is None:
- n_sample_points = 4 * self.get_num_curves() + 1
- points = np.array(
- [self.point_from_proportion(a) for a in np.linspace(0, 1, n_sample_points)]
+ def get_arc_length(self, sample_points_per_curve: Optional[int] = None) -> float:
+ """Return the approximated length of the whole curve.
+
+ Parameters
+ ----------
+ sample_points_per_curve
+ Number of sample points per curve used to approximate the length. More points result in a better approximation.
+
+ Returns
+ -------
+ float
+ The length of the :class:`OpenGLVMobject`.
+ """
+
+ return np.sum(
+ length
+ for _, length in self.get_curve_functions_with_lengths(
+ sample_points=sample_points_per_curve
+ )
)
- diffs = points[1:] - points[:-1]
- norms = np.array([get_norm(d) for d in diffs])
- return norms.sum()
def get_area_vector(self):
# Returns a vector whose length is the area bound by
diff --git a/manim/mobject/types/vectorized_mobject.py b/manim/mobject/types/vectorized_mobject.py
index 6a5e9f79e..80f9e33b1 100644
--- a/manim/mobject/types/vectorized_mobject.py
+++ b/manim/mobject/types/vectorized_mobject.py
@@ -13,7 +13,7 @@ __all__ = [
import itertools as it
import sys
-from typing import Iterable
+from typing import Iterable, Optional
import colour
@@ -896,7 +896,7 @@ class VMobject(Mobject):
return bezier(self.get_nth_curve_points(n))
def get_nth_curve_function_with_length(
- self, n: int, n_sample_points: int = 10
+ self, n: int, sample_points: Optional[int] = None
) -> typing.Tuple[typing.Callable[[float], np.ndarray], float]:
"""Returns the expression of the nth curve along with its (approximate) length.
@@ -904,7 +904,7 @@ class VMobject(Mobject):
----------
n
The index of the desired curve.
- n_sample_points
+ sample_points
The number of points to sample to find the length.
Returns
@@ -915,9 +915,12 @@ class VMobject(Mobject):
The length of the nth curve.
"""
+ if sample_points is None:
+ sample_points = 10
+
curve = self.get_nth_curve_function(n)
- points = np.array([curve(a) for a in np.linspace(0, 1, n_sample_points)])
+ points = np.array([curve(a) for a in np.linspace(0, 1, sample_points)])
diffs = points[1:] - points[:-1]
norms = np.apply_along_axis(get_norm, 1, diffs)
@@ -953,10 +956,15 @@ class VMobject(Mobject):
yield self.get_nth_curve_function(n)
def get_curve_functions_with_lengths(
- self,
+ self, **kwargs
) -> typing.Iterable[typing.Tuple[typing.Callable[[float], np.ndarray], float]]:
"""Gets the functions and lengths of the curves for the mobject.
+ Parameters
+ ----------
+ **kwargs
+ The keyword arguments passed to :meth:`get_nth_curve_function_with_length`
+
Returns
-------
typing.Iterable[typing.Tuple[typing.Callable[[float], np.ndarray], float]]
@@ -966,7 +974,7 @@ class VMobject(Mobject):
num_curves = self.get_num_curves()
for n in range(num_curves):
- yield self.get_nth_curve_function_with_length(n)
+ yield self.get_nth_curve_function_with_length(n, **kwargs)
def point_from_proportion(self, alpha: float) -> np.ndarray:
"""Get the bezier curve evaluated at a position P,
@@ -983,13 +991,15 @@ class VMobject(Mobject):
Point evaluated.
"""
- curves_with_lengths = list(self.get_curve_functions_with_lengths())
+ if alpha == 1:
+ return self.get_points()[-1]
- total_length = np.sum(length for _, length in curves_with_lengths)
- target_length = alpha * total_length
+ curves_and_lengths = tuple(self.get_curve_functions_with_lengths())
+
+ target_length = alpha * np.sum(length for _, length in curves_and_lengths)
current_length = 0
- for curve, length in curves_with_lengths:
+ for curve, length in curves_and_lengths:
if current_length + length >= target_length:
if length != 0:
residue = (target_length - current_length) / length
@@ -1060,15 +1070,26 @@ class VMobject(Mobject):
# Probably returns all anchors, but this is weird regarding the name of the method.
return np.array(list(it.chain(*[sm.get_anchors() for sm in self.get_family()])))
- def get_arc_length(self, n_sample_points=None):
- if n_sample_points is None:
- n_sample_points = 4 * self.get_num_curves() + 1
- points = np.array(
- [self.point_from_proportion(a) for a in np.linspace(0, 1, n_sample_points)]
+ def get_arc_length(self, sample_points_per_curve: Optional[int] = None) -> float:
+ """Return the approximated length of the whole curve.
+
+ Parameters
+ ----------
+ sample_points_per_curve
+ Number of sample points per curve used to approximate the length. More points result in a better approximation.
+
+ Returns
+ -------
+ float
+ The length of the :class:`VMobject`.
+ """
+
+ return np.sum(
+ length
+ for _, length in self.get_curve_functions_with_lengths(
+ sample_points=sample_points_per_curve
+ )
)
- diffs = points[1:] - points[:-1]
- norms = np.apply_along_axis(get_norm, 1, diffs)
- return np.sum(norms)
# Alignment
def align_points(self, vmobject):