Head-locked content stays in the same place in the view

What it is

A battery percentage pinned to the upper-right of the view stays in that corner when the head turns to the ceiling, and still in that corner when the wearer looks down at their shoes. That is head-locked content — also called view-locked, or a HUD. The pose is written in the headset camera frame. Its place on the retina does not turn with the room and does not turn with the torso. There is no “look away to stop seeing it”: what turns away is the world, not this element.

Why it happens

Head-lock parents the interface to the camera: the model matrix is constant in view space. Each IMU sample updates headset pose, the world streams through the frustum, and the locked content’s projected coordinates do not change. It cannot be walked away from in the room, and a head turn cannot slide it out of the field of view. It occupies view coordinates, not environment coordinates.

“Same place” means the same place on the display plane, not a fixed world depth. The same HUD can still be drawn at fifty centimetres or two metres of virtual depth; the depth ray simply always starts between the eyes and aims at a fixed pixel in the view. Near, it occludes more of the world; far, it competes with distant scenery for vergence — either way, it follows the head.

Studying it

During free head turning, have people report a fixed piece of status (battery, reticle, a loading word). The operational definition of head-lock is: whatever the head does in yaw and pitch, that status keeps the same screen coordinates. Re-parent the same status to the torso or the world and those coordinates should move with the head.

Independent variables: parent node (camera / torso / world), eccentricity of the status in the field of view. Dependent variables: pixel displacement of the target on the frame, number of times the status leaves the view during head turns (should be zero for head-lock), search time to find the status.

Asking “is it always there” will also count a body-locked chest panel. The question is whether a head turn leaves it in the same corner of the picture, not whether the person can still find it.

Where it stops holding

Foveated UI that pins content to the current gaze is an eye-movement frame, not head-lock. On a 3DoF device, head-lock and “follows rotation” look similar, because there is no translation with which to split camera from world. In video see-through, a head-locked reticle drawn about the wrong camera centre has parallax against the real world and reads as a film floating in air — yet it still satisfies “screen coordinates do not change.” A half-second “Saved” toast on the head is over before anyone turns; the three lock modes have not had time to behave differently.

Applying it

• Head-lock only the status that must be readable under any heading: a reticle, a recording indicator, a tracking-lost mark. Do not pin a paragraph that needs to be read into view coordinates.
• Keep head-locked elements small, eccentric, and ignorable: off the view centre, small enough that the periphery can sense them and a fixation can finish one line.
• Do not treat head-lock as the default that saves you from placing a world pose. Default by asking whether the information belongs to the room or to the body.
• How to check: record a first-person clip of casual yaw, nodding, looking up. Mark the element’s screen position frame by frame; during yaw and pitch it should sit on the same screen coordinates. If it walks around the picture, the parent is not the camera.

Related

• Same group: N3.01.1 World-locked content stays fixed in the environment · N3.01.2 Body-locked content moves with the person but not with head turn · N3.01.4 Head-lock is the most uncomfortable mode for prolonged use
• Nearby: N3.08 World-locked, Body-locked, and Head-locked · N2.07 The Body as Reference Frame
• Search terms: head-locked · view-locked HUD · screen-space UI