It is a fast TypeScript library for detecting collisions between bodies: Points, Lines, Boxes, Polygons, Ellipses and Circles, and for Raycasting. All bodies can have offset, rotation, scale, bounding box padding, be static (non moving) or be trigger bodies (non colliding).
This library combines:
- efficiency of Bounding Volume Hierarchy (BVH) for broad-phase searching and
- accuracy of Separating Axis Theorem (SAT) for narrow-phase collision testing.
$ npm install detect-collisions
https://prozi.github.io/detect-collisions/modules.html
System
extends RBush
so it has all of its functionalities.
To start, create a unique collisions system:
const physics = new System();
setPosition(x, y)
method - calling it updates bounding boxx & y
properties - setting any of those updates bounding boxpos.x, pos.y
properties - setting those doesn't update bounding boxangle
property andsetAngle()
method - to rotatescale
property andsetScale()
method - to scale (forCircle
takes 1 parameter,x, y
for rest)offset
andsetOffset()
method - for offset from center of bodygetAABBAsBBox()
method - for getting bbox even on non inserted bodiescenter()
method - for centering anchor (useless but available forCircle, Ellipse
)isStatic
property - if true body doesn't moveisTrigger
property - if true body doesn't trigger collisionsisCentered
property - if true the body is centered (true forCircle, Ellipse
)isConvex
property - if true the body is convex (may be false only forPolygon
)padding
property - ignores costly tree update until outside bbox with padding
Box
haswidth & height
properties
bbox = { minX, minY, maxX, maxY }
property - without paddingminX, minY, maxX, maxY
properties - bbox plus paddingsystem
property - to usebody.system.updateBody(body)
internally
- Circle - Shape with infinite sides equidistant of radius from its center position
- Ellipse - Flattened circle (implemented as polygon)
- Polygon - Shape made up of finite number of line segments
- Box - Rectangle (implemented as polygon)
- Line - Line (implemented as 2-point polygon)
- Point - A single coordinate (implemented as tiny box)
Last optional parameter for body creation is always BodyOptions
const options = {
angle: 0,
center: false,
isStatic: false,
isTrigger: false,
padding: 0;
}
// create with options, without insert
const circle = new Circle(position, radius, options);
const polygon = new Polygon(position, points, options);
// insert, without create
physics.insert(circle);
physics.insert(polygon);
// create with options, and insert
const circle = physics.createCircle(position, radius, options);
const polygon = physics.createPolygon(position, points, options);
setPosition
: this modifies the element.pos.x
and element.pos.y
and updates its bounding box in collision physics.
circle.setPosition(x, y);
polygon.setPosition(x, y);
physics.remove(circle);
physics.remove(polygon);
-
After body moves, its bounding box in collision tree needs to be updated.
-
This is done under-the-hood automatically when you use setPosition().
Collisions systems need to be updated when the bodies within them change. This includes when bodies are inserted, removed, or when their properties change (e.g. position, angle, scaling, etc.). Updating a collision system can be done by calling update()
which should typically occur once per frame. Updating the System
by after each position change is required for System
to detect BVH
correctly.
// update one body, use anytime
physics.updateBody(body);
// update all bodies (use 0-1 times per frame):
physics.update();
The preferred method is once-in-a-gameloop checkAll and then handler:
physics.checkAll(handleCollisions);
If you really need to check one body then use:
physics.checkOne(body, handleCollisions);
It is possible to skip the broad-phase search entirely and call checkCollision()
directly on two bodies.
if (physics.checkCollision(polygon, line)) {
console.log("Collision detected!", physics.response);
}
There is often a need for detailed information about a collision in order to react to it appropriately. This information is stored inside physics.response
object. The Response
(documentation) object has several properties set on them when a collision occurs:
a
- The first object in the collision.b
- The second object in the collison.overlap
- Magnitude of the overlap on the shortest colliding axis.overlapN
- The shortest colliding axis (unit-vector)overlapV
- The overlap vector (i.e. overlapN.scale(overlap, overlap)). If this vector is subtracted from the position of a, a and b will no longer be colliding.aInB
- Whether the first object is completely inside the second.bInA
- Whether the second object is completely inside the first.
A common use-case in collision detection is negating overlap when a collision occurs (such as when a player hits a wall). This can be done using the collision information in a Response
object (see Getting Detailed Collision Information).
The three most useful properties on a Response
object are overlapV
, a
, and b
. Together, these values describe how much and in what direction the source body is overlapping the target body. More specifically, overlapV.x
and overlapV.y
describe the scaled direction vector. If this vector is subtracted from the position of a, a and b will no longer be colliding.
These values can be used to "push" one body out of another using the minimum distance required. More simply, subtracting this vector from the source body's position will cause the bodies to no longer collide. Here's an example:
if (physics.checkCollision(player, wall)) {
const { overlapV } = physics.response;
player.setPosition(player.x - overlapV.x, player.y - overlapV.y);
}
// create self-destructing collider
const testCollision = ({ x, y }, radius = 10) => {
const circle = physics.createCircle({ x, y }, radius);
const potentials = physics.getPotentials(circle);
const collided = potentials.some((body) =>
physics.checkCollision(circle, body)
);
physics.remove(circle);
return collided;
};
Hollow / non-convex polygons are fully supported since v6.3.0
the System.response.aInB
and System.response.bInA
is currently because of complexity and speed reasons only checking the bounding boxes of compared bodies. for more informations relate to this issue on github and this merged pull request
For debugging, it is often useful to be able to visualize the collision bodies. All of the bodies in a Collision system can be drawn to a <canvas>
element by calling draw()
and passing in the canvas' 2D context.
const canvas = document.createElement("canvas");
const context = canvas.getContext("2d");
context.strokeStyle = "#FFFFFF";
context.beginPath();
physics.draw(context);
context.stroke();
Bodies can be individually drawn as well.
context.strokeStyle = "#FFFFFF";
context.beginPath();
polygon.draw(context);
circle.draw(context);
context.stroke();
The BVH can also be drawn to help test Bounding Volume Hierarchy.
context.strokeStyle = "#FFFFFF";
context.beginPath();
physics.drawBVH(context);
context.stroke();
Some projects may only have a need to perform SAT collision tests without broad-phase searching. This can be achieved by avoiding collision systems altogether and only using the checkCollision()
function. Note that unless a use-case really requires this, I strongly advise to use the normal flow.
const circle = new Circle(position, radius);
const polygon = new Polygon(position, points);
if (physics.checkCollision(polygon, circle)) {
console.log(physics.response);
}
To get raycast information use
const start = { x: 0, y: 0 };
const end = { x: 0, y: -10 };
const hit = physics.raycast(start, end);
if (hit) {
const { point, collider } = hit;
console.log({ point, collider });
}
- point is the
Vector { x, y }
with coordinates of (closest) intersection - collider is the reference to body of the (closest) collider
Projects requiring physics are encouraged to use one of the several physics engines out there (e.g. Matter.js, Planck.js). However, many projects end up using physics engines solely for collision detection, and developers often find themselves having to work around some of the assumptions that these engines make (gravity, velocity, friction, etc.). Detect-Collisions was created to provide robust collision detection and nothing more. In fact, a physics engine could easily be written with Detect-Collisions at its core.
This isn't caused by faulty collisions, but rather how a project handles its collision responses. There are several ways to go about responding to collisions, the most common of which is to loop through all bodies, find their potential collisions, and negate any overlaps that are found one at a time. Since the overlaps are negated one at a time, the last negation takes precedence and can cause the body to be pushed into another body.
One workaround is to resolve each collision, update the collision system, and repeat until no collisions are found. Keep in mind that this can potentially lead to infinite loops if the two colliding bodies equally negate each other. Another solution is to collect all overlaps and combine them into a single resultant vector and then push the body out, but this can get rather complicated.
There is no perfect solution. How collisions are handled depends on the project.
$ git clone https://github.com/Prozi/detect-collisions.git
$ cd detect-collisions
$ yarn
$ yarn benchmark [milliseconds=1000]
will show you the Stress Demo results without drawing, only using Detect-Collisions and with different N amounts of dynamic, moving bodies
typical output:
βββββββββββ¬ββββββββ¬ββββββββββββββββββββββ
β (index) β value β name β
βββββββββββΌββββββββΌββββββββββββββββββββββ€
β 0 β 317 β 'Total FPS' β
β 1 β 124 β 'FPS / 1000 items' β
β 2 β 57 β 'FPS / 2000 items' β
β 3 β 34 β 'FPS / 3000 items' β
β 4 β 24 β 'FPS / 4000 items' β
β 5 β 22 β 'FPS / 5000 items' β
β 6 β 15 β 'FPS / 6000 items' β
β 7 β 13 β 'FPS / 7000 items' β
β 8 β 10 β 'FPS / 8000 items' β
β 9 β 9 β 'FPS / 9000 items' β
β 10 β 9 β 'FPS / 10000 items' β
βββββββββββ΄ββββββββ΄ββββββββββββββββββββββ
Done in 14.58s.