3D Selection Vault

FocusFlow

Also known as Virtual Window, Gaze-depth interaction

The user looks at an object and then deliberately refocuses the eyes on a nearer, invisible plane; the change in focus depth confirms the selection and opens a virtual window for that object.

How it works

Binocular gaze rays are projected onto a horizontal plane to estimate focus depth, smoothed with a 0.2 s moving average. Gaze direction picks the target object by ray intersection; when the estimated depth then falls into an activation zone around a hidden Virtual Window placed within 2 m of the user (where no scene objects are), the window activates (e.g. a preview, toolbar or zoom lens) and looking back out to the scene closes it. Visual cues teach the depth shift: strong (central circle), weak (corner frame) or adaptive (a circle whose opacity grows as focus moves closer, with a shrinking range during training). In a study (N=24, HTC Vive Pro Eye) with paintings 8 m away, adaptive learning let all participants activate without cues; activation took about 1.3 s, with fewer false triggers than 0.5 s dwell.

Papers