Optics

Pinhole camera frusta and stereo rigs.

import { ... } from 'cseen/optics';

Functions

frustum

A pinhole camera with its image plane and rays.

function frustum(o?: FrustumOpts): Frustum

Parameter	Type
o?	FrustumOpts

Returns Frustum

Options: FrustumOpts

Example

import { Scene, box, ground } from 'cseen';
import { frustum } from 'cseen/optics';

const s = new Scene({ view: { yaw: 62, pitch: 22 } });
s.add(ground(), box({ at: [1.4, 5, 0], size: [1, 1, 1.4], role: 'faint' }));
s.add(frustum({ at: [-1, -1, 1.6], target: [0.4, 5, 0.6], f: 1.6, fov: 46, depth: 4.6, labels: true }));

Run it in the playground →

stereoRig

Two cameras looking at one point: the picture behind stereo vision and depth from disparity.

function stereoRig(o?: StereoOpts): StereoRig

Parameter	Type
o?	StereoOpts

Returns StereoRig

Options: StereoOpts

Example

import { Scene, ground } from 'cseen';
import { stereoRig } from 'cseen/optics';

const s = new Scene({ view: { yaw: 20, pitch: 55 } });
s.add(ground(), stereoRig({ baseline: 2.4, f: 1.5, point: [0.4, 4.2, 0.5] }));

Run it in the playground →

Classes

Frustum

The classic picture of a pinhole camera: the centre, the image plane at focal length f, the rays through its corners and the optical axis. Give it points and it projects them. f and fov animate.

new Frustum(o?: FrustumOpts): Frustum

Options: FrustumOpts

Extends Node.

Member	Description
o: FrustumOpts
centre(): (): V3	the camera centre, as an anchor that follows the node
project(P: V3): V3 | null	where a point (in the node’s space) lands on the image plane, in world space, or null if behind the camera
collect(out: Prim[]): void	appends what this node draws right now to out, in world coordinates. Draws nothing when fully transparent.

StereoRig

Two parallel cameras a baseline apart looking at one point: both image planes, both rays, the horizontal epipolar lines, and the depth Z = f * b / d. baseline, f, fov and point animate, so you can watch the disparity shrink as the point moves away.

new StereoRig(o?: StereoOpts): StereoRig

Options: StereoOpts

Extends Node.

Member	Description
o: StereoOpts
disparity(): number	disparity: how far apart the two image points are along x, in world units. The depth is f * b / disparity.
depth(): number	the depth recovered from the disparity, f * b / d. It equals the point’s distance from the baseline.
collect(out: Prim[]): void	appends what this node draws right now to out, in world coordinates. Draws nothing when fully transparent.

Options and types

FrustumOpts

Options of frustum.

Extends NodeOpts.

Option	Type	Default	Description
target?	V3		the point the camera looks at, in the node’s space. Default: straight ahead along +y.
f?	number		focal length: how far the image plane is from the camera centre. Animatable.
fov?	number		vertical field of view in degrees. Sets the size of the image plane. Animatable.
aspect?	number		width over height of the image plane
depth?	number		how far the rays reach
points?	Point[]		points to project. A plain [x, y, z] is in the camera’s own space; a node or an anchor is wherever it is in the scene, and follows when either moves. Each gets a ray to the camera and a dot where it lands on the image plane
far?	boolean		draw the far end of the pyramid too
labels?	boolean		write C, f, and the name of the image plane
epipolar?	boolean		draw a horizontal line across the image plane through each projected point (the epipolar line of a rectified pair)
roll?	number		tilt the camera about its own axis, in degrees
role?	string		colour role (ink, faint, accent, s1 to s8) or a role you define in CSS

StereoOpts

Options of stereoRig.

Extends NodeOpts.

Option	Type	Default	Description
baseline?	number		distance between the two cameras. Animatable.
f?	number		focal length of both cameras
fov?	number		vertical field of view in degrees
aspect?	number		width over height of the image planes
point?	V3		the point both cameras see, in the rig’s space. Animatable.
depth?	number		how far the rays reach
labels?	boolean		write the names of the parts
role?	string		colour role (ink, faint, accent, s1 to s8) or a role you define in CSS