A ray turns distant targets into an angular selection
Aliases: ray selection · angular pointing · raycast pointing
What it is
A painting on the far wall, a switch three metres off, a panel at the hip: once a ray leaves the hand, they become the same problem — which object does the ray hit. Ray pointing rewrites distant targets as angular selection. What the person controls is not the Euclidean distance from hand to object, but the direction the ray sweeps. Near and far collapse onto one angular quantity.
This is not “reach over there.” Reaching requires the object to be in range. A ray requires only a direction. The hand can hang at the hip and the target can sit across the room; the selection still stands.
Why it happens
A ray originating at the hand or controller drops 3D picking to pointing on a sphere. Intersection walks the line; the first hit interactable is selected. Physical size therefore drops out of the control loop. What enters is the angle the object subtends at the ray origin. A 10 cm panel at 50 cm and the same panel at 5 m are two different targets to the ray — large angle near, small angle far — even though the finger is still “turning a direction.”
That reduction is why the ray exists. Far-field objects cannot be touched, yet they can be hit by a line, so spatial interaction need not pile every control inside arm’s radius. The cost is in the same trade: the control variable changes from “get there” to “hit that angle.” Error and difficulty from here on are accounted in degrees, not in metres.
Studying it
Compare ray selection with direct reaching across distances, and test whether the task has in fact become angular.
Independent variables: target distance, physical size (hence subtended angle), ray origin (hand / eye / shoulder). Dependent variables: selection time, whether the first hit is correct, hand displacement amplitude, reported strategy (“I am lining up” versus “I am reaching”).
If distance changes while subtended angle is held constant (the object scales with distance), ray scores should stay roughly flat, while direct reach collapses beyond arm length. That is a clean test that the conversion to angular selection has happened. Origin at the eye versus the hand changes the angle of the same target; methods must state which, not mix them in one report.
Where it stops holding
Bent rays, teleport parabolas, or “snap to nearest object” are no longer pure angular selection; gravity or magnetism enters the hit rule, and lining up the angle is no longer sufficient. On objects close enough to touch, people abandon the ray for the hand and the conversion does not occur. Gaze selection is also angular, but the origin sits in the orbit rather than the hand, so the tremor and fatigue sources differ. Labs often assume an infinitely long, zero-width ray; consumer stacks have a max length, collision layers, and a thickened UI pick volume, and behaviour drifts from “a geometric line.”
Applying it
- Size and place far-field controls in subtended angle, not as centimetres hard-coded in world space.
- Make the ray’s visualisation (line, hit spark, cursor at the landing) name the currently intersected object, because selection has left the location of the hand.
- Do not force near, grabbable objects through a ray; leave a reach path so a spatial task is not crushed into an angular one.
- How to check: place the same control at 1 m and at 4 m, physical size unchanged, and time selection. If time worsens sharply with distance, you are using a ray while still thinking in “get there.” To make both stations equally usable, the far one must grow with angle.