2D and 3D geometry in Python. This library was originally a part of the DeepDRR package, but has been split out into its own package for easier use in other projects. It contains a lot of custom implementations with no overarching design philosophy, other than getting me through my PhD.
The naming conventions and general approach to coordinate transforms adheres to this tutorial.
It is written solely in Python, and so is not particularly fast. It is also not particularly well-tested, so use at your own risk.
pip install killeengeoimport killeengeo as kg
# Define some points in a world coordinate frame
a = kg.point(1, 2, 3) # a point in 3D
b = kg.point(4, 5, 6) # another point in 3D
v = kg.vector(7, 8, 9) # vector
l = kg.line(a, b) # line connecting a and b
pl = kg.plane(a, v) # plane containing a, b, and normal v
# Also supported: line segments, rays
# Define some coordinate frames
world_from_A = kg.FrameTransform.from_translation(a) # coordinate frame with origin at a
world_from_B = kg.FrameTransform.from_translation(b) # coordinate frame with origin at b
# Transforms to obtain points in those frames
B_from_world = world_from_B.inverse() # transform from B to world
A_from_world = world_from_A.inverse() # transform to A from world
# Transform between the new frames
A_from_B = A_from_world @ world_from_B # transform to B from A
# Objects in the new frame
a_in_A = A_from_world @ a
v_in_A = A_from_world @ v
l_in_A = A_from_world @ l
pl_in_A = A_from_world @ plMost of the time kg objects can be treated like numpy arrays, and they are converted to the expected form.
import numpy as np
points = np.array([a, b]) # [2, 3] array where each row is a point
B2A = np.array(A_from_B) # [4, 4] array representing the transform from B to AAbove is the only use of the X2Y naming convention for frame transformations present here. For more information on why this variable naming convention is awful, see this tutorial.
Documentation is (maybe) available at killeengeo.readthedocs.io.