The reference run for a public event

Every public event has one reference run. That run defines the track for the whole event: where the start is, where the finish is, and which intermediate splits the leaderboard ranks against. Every other run on the event is aligned to this one, and every time on the leaderboard is measured against it.

This page is for organizers choosing the reference run. The mechanics of reference runs in general are covered in Reference run (concept); this page is about the choices that matter at a public event.

What "defines the track" actually means

When a rider uploads a run on your event, Racecraft aligns it to your reference run and predicts the splits at the same checkpoints. That means:

• the start of every rider's clock is the start frame of the reference run
• the finish is the finish frame of the reference run
• the intermediate split times are at the same points the reference's splits sit

If your reference's start is half a second before the actual start gate, every rider on the leaderboard has half a second of un-pedalled time at the front of their run. The reference's splits are the ground truth.

Picking the right reference

The strongest reference for a public event is:

• a complete pass from intended start to intended finish, no deviation
• a clean run — no crashes, no walking, no obvious mistakes
• clear footage — stable mount, no fog, no obstruction
• representative of the line riders should take — not a wild alternate line

It does not need to be the fastest run. The reference defines geometry; the leaderboard captures speed.

In practice, organizers use one of three patterns:

1. A pre-recorded run by the organizer or a designated rider the day before, on the actual race line. This is the cleanest version — you control the camera setup and the line.
2. The track-builder's or course-designer's run if they're filming. Often the most authoritative on the intended line.
3. A confident rider's first qualifying run if there isn't time to pre-record. Workable, but tighter on the camera-setup constraints — see below.

Camera setup matters more on a public event

For private analysis, riders often use one camera setup all season. On a public event, the reference run was probably filmed by one person, and every other rider's footage was filmed differently — different cameras, different mounts, different fields of view.

Alignment still works across this, but it works better when the reference run uses a setup riders can match:

• helmet-mounted, chin-up is the most common rider setup. A reference run from this same angle aligns most reliably.
• avoid unusual angles for the reference — chest mounts, bike- frame mounts, follow-cam — even if you have nice footage from one. Use the closest-to-typical view as the reference and save the unusual angle for highlight reels.

Splits on the reference run

Splits are the checkpoints riders are measured against. Place them on visually distinctive frames the alignment can lock onto reliably:

• the start as close to the actual start gate as possible
• the finish at the actual finish line
• intermediate splits at meaningful section boundaries — the entry to a tech section, a notable corner, a transition between parts of the track

Three to six intermediate splits is the common range. Fewer and you lose section-level visibility on the leaderboard; more and the leaderboard becomes noisy without telling you more.

We confirm and lock in the splits as part of setting up the event — if you have preferences for where the splits should sit, send them with your event request.

Changing the reference mid-event

You can change the reference run during a live event. When you do, the splits on every existing rider's run are re-predicted against the new reference, and the leaderboard re-ranks against the new template.

Common reasons to change mid-event:

• you got cleaner footage on the morning of the race
• the original reference had a camera issue that was hurting alignment for everyone
• the track was modified between practice and race day

To swap a reference mid-event, contact us — we'll re-point the reference and the leaderboard updates.

Common pitfalls

• using a fast-but-messy run as the reference — the splits get placed on bad frames and every rider's alignment suffers
• using footage with the camera knocked sideways — a tilted reference makes target alignment less reliable
• using a multi-lap or warmup-included file — the reference must be one clean lap, start to finish
• picking the reference's splits at non-distinctive points — long featureless stretches don't anchor well; corner entries and trail markers do