Larger travel needs a longer duration

What it is

The same clock covering 24 pixels and covering a full screen differs by an order of magnitude in mean speed. Long duration on a short hop looks sticky; short duration on a long hop looks like a teleport. Duration has to follow travel distance because the eye tracks speed, not keyframe percentage. This is not about dropped frames on the main thread: jank is insufficient compute; here the question is whether the path can be followed perceptually.

Why it happens

Smooth pursuit has a speed ceiling. Beyond it, gaze switches to a string of saccades and the path subjectively breaks into hops. Hold duration fixed while increasing distance and speed rises linearly until it leaves the trackable band; apparent motion collapses to the two end states. Conversely, a tiny distance with a long duration is so slow it no longer looks like going somewhere — the object feels glued, and the action loop is stretched for no spatial reason.

A second layer: large travel sweeps more background and more intervening objects, so more time is needed to read occlusions and layering along the way. A shared element flying from a list to full screen is not just Euclidean distance between two anchors; scale change adds spatial travel of its own.

Studying it

Hold easing fixed, vary distance and duration systematically, and collect a three-way judgment: teleport / trackable travel / too slow. Eye tracking can separate smooth pursuit from saccade sequences. Independent variables: pixel distance (or visual angle), duration, size change of the object. Dependent variables: judgment category, pursuit gain, accuracy of reported direction.

Prefer visual angle over raw pixels: 400 px is a different angular speed on a phone and a desktop. Report viewing distance.

Where it stops holding

• Cross-fades have almost no spatial travel, so the distance–duration link is weak; this constraint is mainly for displacement and scale.
• If the user is not looking, no speed will build a path; they return to an end state.
• Lengthening duration to keep a large hop trackable can start blocking chained taps. The task has to pick a side: navigational demonstration wants trackability; high-frequency switching wants short.

Applying it

• Bucket duration by distance: short for nearby hops, longer for cross-screen travel. Do not stamp one millisecond value on everything.
• Estimate scale travel from diagonal or area change, not only from the anchor’s translation.
• Check: on the target device at a normal holding distance, compare the farthest and nearest transitions with a direction-report task. Nearby judged sluggish and far judged teleport means speed was not calibrated to distance.

Related

• Same group: F7.02.1 Motion that is too short is not perceived as a transition · F7.02.2 Motion that runs too long blocks the next action
• Nearby: A1.22 Types of eye movement · F7.04 Shared-element transitions · F7.12 Performance cost of motion
• Search terms: travel distance · smooth pursuit · motion duration