Skip to content

Launch-speed estimate — v0.5.76 published source QA and v0.5.77 development

LateBrake adds a session-only 0–5 km/h forward launch estimate measured using the authoritative time since GO. This is not a human reaction-time measurement, a stopwatch-grade crossing timestamp, or a competitive result. No credits, career awards, best times, saves or vehicle handling depend on it.

Rules

  • A real, qualified 3 / 2 / 1 / GO is mandatory; the first valid post-GO chassis sample must be below 5 km/h. A first sample already at or above 5 km/h invalidates the entire attempt rather than allowing a later deceleration to masquerade as the initial launch. An observed below 5 km/h chassis-forward sample must precede an observed 5 km/h or more forward sample within 30 seconds of GO.
  • Source input is the genuine player's signed forward Chaos velocity, not absolute speed, throttle command, guessed acceleration or rival position.
  • Use active race clock minus charged recovery penalties. Invalid frames, nonfinite velocity, red lights, pause/menu/focus loss and unqualified physics cannot invent a launch. A clock rollback or over-30s threshold invalidates an uncaptured estimate. A successful paid recovery invalidates an uncaptured launch attempt; a previously captured estimate remains unchanged. Failed recovery requests cannot erase it.
  • An accepted restart clears prior telemetry. A captured observation is immutable for the remaining race. No persistent records are written.
  • Expanded HUD shows an approximate launch time near the start; Results optionally shows the same session estimate. Small screens can clip optional telemetry after authoritative time/position readouts.

Merged source refinement (v0.5.77, unreleased and not a playable build)

  • For two consecutive valid post-GO physics samples separated by at most 0.25 seconds, estimate the first 5 km/h crossing by bounded linear interpolation. The estimate must lie between the measured samples; it is an approximation, not reaction time, an exact sensor timestamp or performance-certified acceleration.
  • On malformed/nonfinite velocity, pause, focus loss, unqualified frame or a longer data gap, do not interpolate across the interruption. The next physically observed threshold is used conservatively instead.
  • Reject a forward 5 km/h crossing reported at the exact timestamp of the preceding lower-speed sample; the two readings cannot prove elapsed acceleration. An initial already-moving reading still invalidates the launch attempt. Verified recovery and race restart keep their existing safety semantics.
  • The policy is run-local HUD/Results information only. No credits, personal bests, career progression, checkpoint gates or tuning depend on the interpolated value.

Automated source checks

cmake -S Tests/Standalone -B Tests/Standalone/build
cmake --build Tests/Standalone/build --config Release
ctest --test-dir Tests/Standalone/build -C Release --output-on-failure -R LaunchSpeedMeter
python -m unittest discover -s Tests/Editor -p LaunchSpeedBindingTests.py -v

Still mandatory on an authorized Unreal Engine 5.8 installation

  • Check UHT/Editor/game compilation, actual forward Chaos velocity and actor tick order; verify that a stationary physical grid start really produces the expected first below-threshold sample.
  • Test W/gamepad/touch launch, reverse then drive, 3/2/1/GO, pause, focus loss, 5-second paid recovery, grid restart and repeated quick races.
  • Inspect desktop/mobile HUD overflow and Results detail-page pagination. Compare on-screen estimates with recorded physics logs rather than describing them as human reaction times.
  • Record exact installed Win64/Android package hashes and observed device/OS/frame-time evidence before calling any build playable.

This source feature does not close #148 or #13.