Enter with a fast start and slow settle; leave with a slow start and fast exit
Aliases: decelerate on enter · accelerate on exit · enter–exit curves
What it is
Entry ends in a resting, usable object; exit ends in leaving the field. Those two jobs want opposite velocity profiles. Entry needs to be visible immediately (high speed) then decelerate so the hit target becomes aimable in the second half. Exit needs a readable “this is leaving,” then acceleration out, so it does not linger in the way of the next act. Arrival and departure are asymmetric; they are not a mirrored fade pair.
Why it happens
On enter, the object has to go from absent to tappable. High speed in the first half makes a strong transient so attention can catch it; deceleration in the second half stabilises location so pointing can lock. A slow start on enter leaves the object small and sluggish for a long time, as if the system were not ready.
On exit the object is already present. A slow start lets “it is leaving” be read as an intentional departure rather than a pop-out; acceleration afterward clears pixels and hit area quickly and returns the loop. Decelerating toward off-screen keeps a soon-to-be-useless object in view extra time and blocks the next action.
Neither end wants constant speed: the brake on enter and the start on exit are two different physical stories — come to rest versus leave.
Studying it
Cross two curves on the same enter/exit pair (fast-out-slow / slow-out-fast / linear). Collect estimates of “when did this become tappable” and the wait before the next tap after exit. Independent variables: enter–exit curve pairing, object size. Dependent variables: estimated moment of interactivity, input gap after exit, naturalness.
The objective moment when the hit region is stable and the subjective “it has settled” often diverge; log both.
Where it stops holding
- With a shared element across pages, the enter curve must meet the other page’s exit; talking about “fast then slow” in isolation can be interrupted by that join.
- Under reduced motion, travel is often replaced by a fade, so enter/exit curve asymmetry barely applies; the cross-fade matters more.
- Spring overshoot postpones “settled” until after several oscillations and undoes the benefit of a decelerating arrival.
Applying it
- Panels, menus and cards should enter with high initial speed then brake; they should leave with a readable start then clear the screen quickly.
- Do not stamp the same ease-out curve on both enter and leave.
- Check: allow taps roughly 100 ms before enter finishes and watch miss rate; on exit, measure delay from animation start to the next successful input. If people still will not tap on enter, or still wait on exit, the profiles are reversed.