On 24 March 1877, the Oxford–Cambridge Boat Race ended in a dead heat – the only draw in its history. The finishing judge, a professional waterman nicknamed “Honest” John Phelps, was over 70 and reportedly blind in one eye. Oxford swore they’d won “by feet,” but without evidence, the verdict stood as a tie. The backlash forced rule changes the very next years fixed finishing posts and former “Blues” as umpires – because everyone realized tight finishes needed optics, not opinions.
Think of that scene as sport’s “Zapruder film”: lots of eyes, zero proof. From that seed grew an obsession – build a machine that sees the truth when humans can’t.
Who actually built the solution? (Spoiler: not just one person)
Early experiments existed. In 1890, photographer John C. Hemment rigged a thread across the track to trip a single exposure 0 ingenious, but fragile, and it often missed the decisive instant or couldn’t sort places behind the winner.
By the 1930s, with betting money and reputations on the line, engineers iterated fast:
- 1932, Los Angeles Olympics: The Kirby system paired finish-line imagery with a time indicator in the same frame – primitive by today’s standards, but it set the template: picture + timestamp = evidence.
- 1937, Del Mar (California): Paramount engineer Lorenzo Del Riccio unveiled the Photo-Chart slit/strip camera on Del Mar’s opening day. It recorded only the finish plane and stitched time slices into a single decisive image – the true ancestor of modern line-scan systems.
- 1948, St. Moritz & London Olympics: Swiss timer OMEGA introduced the “Magic Eye” photoelectric cells and paired them with a British Race Finish Recording Company slit camera at the London Games. That combo-electronic triggering + slit image became the world standard after it separated Harrison Dillard from Barney Ewell in the 100 m final.
Fast-forward: at Paris 2024, OMEGA’s Scan’O’Vision Ultimate captured up to 40,000 images per second so judges could split finishes separated by millimeters.
What a photo-finish image really is (and why it looks “wrong”)
A photo-finish picture is not a normal photograph of a moment. It’s a measurement of a place (the finish line) across time.
- The camera points straight down the finish line and records a 1-pixel-wide vertical slit over and over thousands of times per second. Each new “line” is stacked beside the previous one, so left-to-right = time, not track position. That’s why backgrounds look like streaks, wheels look stretched, and athletes appear “sliced.”
- Classic film systems moved a narrow slit of film at about the racers’ speed so only moving subjects appeared; static objects smeared. Modern digital line-scan cameras do the same thing electronically.
TL;DR: You’re looking at the finish line over time, not the whole track in a single instant.
How officials “read” the image
A thin cursor (the “hairline”) is placed on the relevant body part in the stacked image to get the official time. World Athletics recommends scanning ≥1000 lines/sec (often 2000+) and locking alignment to the leading edge of the line so both primary and backup cameras match to 0.001 s.
What body part actually “counts”?
Athletics (track & field): The winner is the first torso (shoulders/chest/abdomen) to break the line – not hands, head, or feet. Officials are trained with an explicit torso definition (clavicle to hip line) for placing the cursor. Times may be read to 0.001 s but are rounded up to the next 0.01 s for results.
Wheelchair races: Timing/placing is taken from the center of the leading wheel’s axle.
Horse racing: Judges place the time line on the leading part of the head – practically, the nose – excluding ears or tongue. That’s why “wins by a nose” is literal.
Cycling: Same line-scan principle as track; a popular explainer in pro cycling circles describes modern CCD/CMOS line-scan setups looking exactly down the finish stripe.
Here’s something for the shutterbugs
Line-scan vs. frame-based cameras
Frame cameras capture a 2D frame at intervals; line-scan captures a continuous 1D stream that’s assembled into a “strip” image. The latter avoids motion blur across lanes and guarantees every lane is observed at the same instant the body part hits the line.
Why the picture looks “bent” or “melted”
Because time is the horizontal axis, anything that accelerates/decelerates “stretches” or “squashes.” Wheels become ovals; horses look elongated; backgrounds smear into bands. That’s normal and expected.
Alignment & dual-view
Major meets use two cameras, one each side of the track (ideally independent power/record paths). Some horse-racing posts mount a mirror strip on the far side so one camera can see both near-side and far-side margins cleanly in one image.
Sampling & accuracy
World Athletics guidance: scan ≥1000 lines/sec (preferably 2000), verify zero tests, and ensure both primary and backup cameras agree to 0.001 s. The timing must round up to the next longer hundredth when reported.
State of the art
OMEGA’s Scan’O’Vision family, upgraded for Paris 2024, captures up to 40,000 images/sec, producing composite strips almost instantly for judges. FinishLynx systems (track, cycling, skating, more) are widespread, with line-scan capture and on-screen hairline timing.
Persistent myths to retire
- “Dive with your hands.” Hands don’t count in track; torso does. A mistimed dive can slow your center-mass forward velocity.
- “The photo is distorted, so it’s fake.” The distortions are a feature of line-scan (time on x-axis), not fakery. Judges are trained to read them correctly.
Why photo finish changed sport
After 1877, sport learned a hard truth – “close enough” isn’t good enough when careers and money ride on millimeters. By fusing optics with timekeeping, photo finish turned arguments into evidence. From Hemment’s single-frame tricks to Del Riccio’s strip camera and the Magic Eye’s electronic trigger, the tech’s arc has always bent toward verifiable truth.
Editor’s corner (tips for publishing photo-finish pics)
- Don’t “correct” the look. The stretched wheels and streaked backgrounds are diagnostic. Avoid filters or AI “fixes” that normalize them.
- Preserve resolution horizontally. The decisive differences often live in a few pixels. Keep export widths generous and use lossless/web-p converter when possible.
- Add a caption that explains the axis. A single line: “Horizontal axis = time, vertical = finish line” to turn confusion into comprehension.
- Keep metadata if you can. Split-second provenance matters when images circulate beyond your site.
