screenToWorld

screenToWorld(screenX, screenY, inverseVP, vpX, vpY, vpW, vpH, planeZ?): object

Defined in: ui/util/math.ts:109

Where a screen point lands on the world plane at

Parameters

screenX

number

screenY

number

inverseVP

Float32Array

vpX

number

vpY

number

vpW

number

vpH

number

planeZ?

number = 0

Returns

object

x

x: number

y

y: number

P

planeZ.

Details

A screen point is a RAY, not a position. Orthographically the ray’s x/y are constant along its length, which is why taking z = 0 worked and why every 2D caller can keep calling this unchanged. Under a perspective camera they are not constant, and the same call would answer with wherever the near plane happened to be — off by more the further the content sits from z = 0.

     So the ray is intersected with the plane, which is one implementation
     with the orthographic case as its degenerate form rather than two
     branches that have to agree. `planeZ` defaults to the 2D plane, so a
     caller that has no depth in mind keeps getting the 2D answer.