This origami math library contains some geometry and some linear algebra; as it turns out a lot of time is spent calculating the intersection of shapes. All math functions are accessible to the user, as well as some top-level primitives.
Vectors can be any dimension, although some methods are specifically for 2D or 3D, and some treat a vector like a point.
varvector =ear. vector( 0.5, 0.666)
2D and 3D specific methods often end in numbers like cross2 and cross3.
blue1 = red1.lerp(red2, t)
blue2 = red2.lerp(red3, t)
yellow = blue1.lerp(blue2, t)
// the x component
0] // also, the x component
This vector object is immutable; methods returns a new, transformed vector.
Matrices are powerful representations of geometric transformations. Inside of one matrix can contain instructions for any number of rotations, translations, reflections, scaling, shearing; and it can tell you things like whether a polygon has been flipped over.
A 3x4 matrix encodes 3D space, its columns from left to right:
Matrices are stored column-major, which lines up vector components so each basis vector is contiguous (eg. translation is indices 9, 10, 11).
An example of a matrix in action is a simulation of fold, using a reflection matrix. The algorithm is:
Given a fold line, split the polygon into two polygons. Use the line to construct a reflection matrix and apply the matrix to one of the polygons.
This library makes it as easy as possible to generate a reflection matrix. Any line type (line, ray, segment) can be turned into a reflection matrix.
varsegment =ear. segment(point1, point2) varmatrix =segment. reflectionMatrix()
Unlike vectors and most objects, matrices are mutable; each method modifies the matrix.
These methods will still return this, the modified matrix.
lines extend infinitely in both directions
rays extend infinitely in one direction
line segments are bound by two endpoints
Lines and rays are created by specifying a direction vector and an origin. Segments are initialized with two endpoints.
line( vector, origin)
ray( vector, origin)
segment( point, point)
If you leave out the origin argument it is assumed to be (0, 0, 0).
Also, it may be easier to create lines and rays with the static method: fromPoints.
When a method requires more than one point, points are grouped using arrays.
Lines, rays, and segments all share a common prototype, thus share some properties and methods, like the method nearestPoint.
varpoint =segment. nearestPoint()
All three have an vector and origin property, which is what a line requires for initialization.
varsegment =ear. segment( 1, 0, 2, -1) varline =ear. line(segment)
A segment being converted into a line (infinite length).
One problem with the vector, origin form is uniqueness; the same line can be represented infinite ways. In Origami and Geometric Constructions Robert Lang offers a solution that makes use of one vector and one scalar.is implemented in this library; so, each line also contains these two properties.
Lines can be uniquely defined by a unit normal vector u and the shortest distance to the origin d.
Polygons are created from an ordered set of points which define the boundary. It's possible to create non-convex and self-intersecting polygons.
Many of the methods in this library are built for convex polygons.
Convex hull is a static initializer that can also be used to fix any winding issues in a set of points.
The straight skeleton is an operation on a polygon that creates a flat-foldable crease pattern, and when folded, the boundary becomes collinear.
varhull =ear.polygon. convexHull( points)
rect( width, height)
rect( x, y, width, height)
enclosingRectangle returns an axis-aligned rectangle, nearest returns the nearest point on the boundary, transform takes in a 3x4 matrix.
circle( x, y, radius)
circle( origin, radius)
Remember, objects are immutable. If you call a object's method expecting a change, check the return value for a new modified object.
The epsilon is the small number used to decide whether floating point numbers are equal.
This library intends to be useful at any scale; the sketches here demonstrate this. The epsilons on this page have been increased so large to the point of being visible.
When a method uses an epsilon, the final parameter is the optional epsilon parameter, with a default value of 1e-6.
Is a point collinear to a line (line, ray, segment)?
varpoint =ear. vector() varline =ear. line() point. overlap(line) line. overlap(point)
Endpoints can be treated as inclusive or exclusive, which includes or excludes the area around their endpoints.
varsegment =ear. segment()
inclusive() // or
Vectors, rays, segments, and polygons (rect, circle) can be made inclusive and exclusive.
Inclusivity for polygons includes or excludes the boundary. This is useful when asking if points are inside or outside, and it's important to include or ignore points on the boundary.
varpoly =ear. polygon() polygon. inclusive() polygon. overlap( point)
Polygons boundaries in black, including and excluding the area around them.
Remember the epsilon in these sketches have been heavily magnified. These calculations are typically happening at an imperceptable scale.
overlap( a, b)
The overlap method is accessible from the top level, or as a member function on primitives.
varsegment1 =ear. segment(). inclusive() varsegment2 =ear. segment(). exclusive() segment1. overlap(segment2)
Inclusive and exclusive overlap
Overlap methods are just intersection methods that return a boolean instead of a point. In some cases this calculation is faster.
These are the algorithms that get called behind the curtain. They are available to be called directly, and require more parameters than the other interface.
overlapConvexPolygonPoint( poly, point, func, epsilon) ear.math. overlapConvexPolygons( poly1, poly2, fn_line, fn_point, epsilon) ear.math. overlapLineLine( aVector, aOrigin, bVector, bOrigin, aFunction, bFunction, epsilon) ear.math. overlapLinePoint( vector, origin, point, func, epsilon)
The overlap algorithms
Some of these parameters are domain functions. These functions mark the span of an acceptable result against the vector of the primitive, including determining inclusivity. These functions are also available:
The inclusive and exclusive domain functions for general case, lines, rays, or segments.
intersect( a, b)
The intersect function will accept most combinations of pairs of primitives, like lines, circles, polygons, also available at the top level or as methods on objects.
Intersections involving convex polygons results in 0, 1, or 2 points, they return an array of points.
Intersections limited to lines only result in 0 or 1 point, they do not return an array, just a point or undefined.
Again, this method is calling other methods under the hood. These are available to you as well.
intersectConvexPolygonLine( poly, vector, origin, fn_poly, fn_line, epsilon) ear.math. intersectCircleCircle( c1_radius, c1_origin, c2_radius, c2_origin, epsilon) ear.math. intersectCircleLine( circle_radius, circle_origin, line_vector, line_origin, line_func, epsilon) ear.math. intersectLineLine( aVector, aOrigin, bVector, bOrigin, aFunction, bFunction, epsilon)
Clip is only available on polygons right now, and it only operates on lines, rays, and segments.
Splitting a line by a convex polygon will result in a smaller segment.
Increasing the epsilon on the clip method demonstrates the priorities. In both inclusive cases the resulting segment must not be shorter than 2 * epsilon.
The inclusive result can extend beyond the polygon within one epsilon unit, and the exclusive cases removes parallel edges by testing exclusive polygon-point overlap with the segment's midpoint.
isCounterClockwiseBetween( circle, angle1, angle2)
The space between two vectors creates two interior angles. It's important to distinguish between vectors a and b the clockwise or the counter-clockwise interior angle.
Currently there is no object-style interface to angle calculations, for now you need to call back end math functions directly.
Get the interior angle between two radians/vectors.
clockwiseAngleRadians( angle1, angle2) ear.math. counterClockwiseAngleRadians( angle1, angle2) ear.math. clockwiseAngle2( vector1, vector2) ear.math. counterClockwiseAngle2( vector1, vector2)
When operating on angles, these methods seamlessly wrap around from 2π to 0 and include negative radians as similar to their positive counterparts.
counterClockwiseVectorOrder( ... vectors)
counterClockwiseRadiansOrder( ... angles_in_radians)
Sort an array of vectors or angles, but return the result as an array of indices instead of modifying the array.
The first argument serves as the "origin" angle that other vectors are sorted around". Index 0 will always be 0.
Sorting is required for such simple things as measuring the interior angles between consecutive vectors.
The red line is the first in the vector array; you can tell the algorithm sorts around it as the sector colors shift more when it is moved.
clockwiseBisect2( vector1, vector2) ear.math. counterClockwiseBisect2( vector1, vector2)
A bisection could refer to the smaller or the larger angle; these methods return one solution by clarifying clockwise or counter-clockwise.
Bisecting two lines is effectively origami axiom #3, and results in two solutions unless the lines are parallel.
The parallel case maintains the index position of each line. The lines are sorted based on if the parameter lines are running in the same direction or not (dot product > 0).
For trisections and beyond, use the general subsect methods.
counterClockwiseSubsect2( divisions, vectorA, vectorB) ear.math. counterClockwiseSubsectRadians( divisions, angleA, angleB)