Skip to content

makc/four-points-controls

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

four-points-controls

Allows you to orient an object in 3D using positions of four points of its local XY plane in the camera plane. Get it from NPM or CDN and then just:

const controls = FourPointsControls( camera, renderer.domElement );
controls.add( mesh ); // your 3D object to orient
scene.add( controls );

Demo

https://makc.github.io/four-points-controls/

API

// camera: an active instance of PerspectiveCamera
// element: optional HTMLElement to add pointer listeners to; if
// not specified, no handles are created and you are responsible
// for updating controls.points values
const controls = FourPointsControls( camera, element );

// four points in the camera plane, set by controls UI if element
// was specified, or by you (both x and y should be normalized to
// -1...1 range
controls.points = [...];

// currently active corner, 0...3 (see exact3 comment below)
const index = controls.currentPoint;

// a method of pose calculation (see below for supported values)
controls.method = ...;

Methods

  • FourPointsControls.exact3
    calculates exact fit of three points and minimizes fourth point reprojection error (this is similar to opencv SOLVEPNP_P3P method); three points used are those around the active corner

  • FourPointsControls.exact3.average
    experimental; calculates approximate fit of four points by averaging the results of FourPointsControls.exact3 for every point

  • FourPointsControls.exact4
    default; calculates exact fit of four points, but sacrifices the transformation orthonormality (this works best with flat or near-flat objects)

  • FourPointsControls.exact4.diamond
    experimental; tweaks the results of FourPointsControls.exact4 to equalize the lengths of X and Y axes

  • FourPointsControls.opencv
    calculates approximate fit of four points using opencv SOLVEPNP_IPPE_SQUARE method (you will need to preload ~9 MB opencv.js build to use this option)