Pivot, recorded and compared.

Microsoft’s Pivot control was recorded, measured, and compared with the Flutter implementation. The results show a closer match—not an indistinguishable renderer.

Reference: WP8.1 emulator
OS build: 8.10.12358.0
Evidence: 57 native trials
Hardware: Lumia pending

Watch the comparison

Native is on the left; Flutter is on the right. Each MP4 already contains both panels on the same recorded input timeline, so one play action starts both views. Flutter frames come from the widget tree, not a redrawn native video.

Tap the next headerDownload MP4

Drag and commitDownload MP4

Wrap to the previous pageDownload MP4

The videos are encoded at 30 fps. Missing native capture time is represented by holding the last captured frame; frames are not interpolated or stretched to force agreement.

What changed

Observed behavior	Flutter change
Headers use native spacing and cyclic order.	`WpPivotView` lays out headers from measured widths and reference coordinates.
A drag moves the current page without continuously exposing its neighbor.	Header and body travel are controlled separately instead of by a stock `PageView`.
Selection commits on release with distinct outgoing and incoming phases.	The transition moves the old page out before bringing the selected page in.
A skipped header rearranges before sliding into place.	The selected header becomes the anchor before a one-slot transition.
The threshold-crossing sample is consumed before body travel.	Gesture tracking follows the delivered-sample sequence found in the logs and images.

The original WpPivot header-only API remains available. The full view is an explicit migration and preserves page state, controller selection, keyboard navigation, and reduced motion.

Measured result

Three pilot inputs compare the retained 2.0 baseline with the revised candidate. Position MAE is mean absolute error in the 480 × 800 reference image.

Pilot	Moving-page MAE	Different visible pages
Header tap	313.0 → 32.2 px	22 → 0 of 28
Committed drag	158.6 → 4.7 px	23 → 1 of 46
Previous wrap	305.3 → 16.1 px	27 → 0 of 36

Native observed header positions and Flutter replay positions over time — **Header transition positions**Native observation and Flutter replay

Supported

Original PNGs and image tracking support layout and coarse-motion comparison.
The candidate matches the selection sequence in 21 fresh trials and 21 earlier core trials.
Nine later header and title gesture trials bring the archive to 57 native trials and 3,830 PNGs.

Not proven

Emulator evidence does not validate Lumia hardware behavior.
Capture timing does not establish display latency or a universal flick threshold.
Font and rasterizer differences remain visible.

Runtime

The release web build replayed four scenarios with one warm-up and three measured repetitions. All 12 measured trials selected the expected page.

Engine measurement	Median	95th percentile	Maximum
Build work	3.00ms	5.20ms	11.60ms
Raster work / submission	0.40ms	0.80ms	24.00ms
Total frame span	3.70ms	6.10ms	28.70ms
Post-selection frame interval	16.60ms	17.30ms	29.30ms

Measured release web frame work including one frame over the chosen budget — **Release-web frame work**678 measured engine frames

These numbers support smooth animation on the measured host. They do not measure physical display presentation, native browser touch input, Windows Flutter, mobile performance, or Lumia hardware.

Evidence

Header gestures and runtime release Original Pivot evidence release Protocol, tools, and reproduction commands

The archives include original frames, input traces, manifests, comparisons, metrics, source snapshots, and SHA-256 inventories. Unsuccessful trials and superseded prototypes remain labeled and available.

Next checks

Wider gesture sampling, mobile runtime measurement, and a separately labeled Lumia cross-check remain open work.