Test method · Version 1.0

How to test an automatic camera trigger.

A trigger can notice an event and still miss the photograph. This method separates detection from camera readiness and completed capture, records false positives, and keeps every result tied to the device and conditions that produced it.

Published 15 July 2026No results published yetDesigned for physical iPhones

The principle

Count the chain, not just the flash.

For every planned event, record whether the app detected it, whether it requested a capture, whether the camera completed that capture, and whether the saved image contains the intended moment. A single “worked” score hides where failure occurred.

No benchmark is claimed here. This is a protocol for future testing. It contains no synthetic success rate, no competitor result and no implication that an app matches scientific lightning-imaging equipment.

Protocol

Six steps, recorded every time.

Freeze the setup

Record app build, iPhone model, iOS version, lens, capture format, trigger mode, sensitivity, analysis region, cooldown, burst and power state.

Describe the scene

Record illuminance or a repeatable light description, tripod or handheld use, distance to stimulus, background motion, temperature and thermal state.

Use a controlled stimulus

Prefer a repeatable indoor light pulse, moving target, fixed-level audio playback or measured device movement. Do not begin a safety test by standing in a thunderstorm.

Run enough trials

Use at least 30 planned events per configuration for an exploratory result. Keep warm-up trials separate and preserve misses instead of rerunning them silently.

Label each outcome

Record detection, capture request, completed save, usable frame, false positive, duplicate, readiness failure and any human intervention as separate fields.

Publish the denominator

Report raw counts with the number of planned events and observation time. Include the CSV, excluded trials and exact reason for every exclusion.

Measurements

What the final report should expose.

A result without a denominator, device or configuration is a product anecdote, not a reproducible measurement.

MetricCalculationWhy it matters
Detection rateDetected planned events ÷ all planned eventsMeasures the trigger model before camera-pipeline failures.
Capture completion rateSaved captures ÷ detected planned eventsShows whether detection became a file.
Usable-frame rateFrames containing the intended moment ÷ all planned eventsCaptures the result a photographer actually cares about.
False-positive rateUnplanned triggers ÷ observation timeAllows comparison across sessions of different length.
End-to-end latencyStimulus timestamp to exposure or saved-frame timestampMust state the clock, recording method and uncertainty.
Readiness failuresCount by reason and camera stateSeparates app detection from AVFoundation readiness.

Suggested matrix

Change one important variable at a time.

TriggerControlled stimulusMinimum conditions to disclose
Lightning / brightnessRepeatable short light pulse outside and inside the selected analysis regionPulse duration, relative brightness, ambient level, region and lens
Visual motionTarget moving across a marked path at repeatable speed and sizeDirection, speed, subject size, background motion and distance
SoundFixed audio sample played at a measured distance and safe volumeSound level if available, microphone orientation, ambient noise and keyword use
Device motionRepeatable tilt, shake or impact using a marked movement rangeMount, movement amplitude, duration and device orientation
Stable sceneTripod settling after a defined disturbanceSupport, stabilization, settling threshold, wind and surface vibration

Safety before footage

Do not field-test lightning from an unsafe position.

The National Weather Service states that there is no safe place outside when thunderstorms are in the area. Controlled indoor stimuli should be used for repeatability; any real-weather observation belongs inside a safe building or hard-topped vehicle and must follow local safety guidance.

Go indoorsIf thunder is audible or lightning is visible, stop the outdoor setup and seek a safe shelter.
Do not touch wired equipmentOfficial guidance also warns against using devices connected to electrical outlets during a storm.
Never trade safety for a sampleA missed field trial is data. It is not a reason to remain exposed.

Primary context

Sources and limits.

Independent testing welcome

Publish the misses too.

Reviewers may use or adapt this protocol without promising a favourable result. If you publish a dataset, link the configuration, raw counts and exclusions so another person can repeat the work.