A camera that shoots 40k FPS decided the 100-meter sprint final
Ultra-fast “line scan” cameras that capture 40,000 lines per second decided the men’s 100m final in Paris, prompting technical deep dives into how these systems work and why the entire image is effectively the finish line over time. Commenters explain the one-pixel-wide sensor, specialized LED advertising strip synchronized for the photo finish, memory and data-rate requirements, and the rule that victory is determined by the torso crossing the line. Others question whether such extreme timing precision is meaningful or fair given physical variances in tracks, wind, and reaction times, and compare this approach with alternatives like transponders or coarser timing with more ties.
How the finish camera works
- It’s a line-scan (strip) camera: the sensor is effectively 1 pixel wide and many pixels tall.
- The vertical axis is space across the track; the horizontal axis is time. Each vertical column is the finish-line view at a specific instant.
- Every pixel in the composite image lies on the geometric finish line; runners look “distorted” because different body parts are captured at slightly different times.
- Shadows, shoes, and limbs are recorded as they cross the line, just like torsos. Distortions (e.g., “ski” feet) are expected.
Advertising banner and calibration
- The “banner” with the Olympic rings and brand logo behind the runners is not a normal 2D sign.
- It’s a very narrow, one-pixel-wide vertical LED strip that rapidly cycles columns of the logo.
- To the naked eye it looks like flicker; to the line-scan camera, stretched over time, it becomes a readable banner.
- Commenters note this is both marketing and a quick visual check that camera alignment and timing are correct.
Timing, data, and FPS semantics
- The camera captures around 40,000 “lines per second,” which some argue is a more accurate term than FPS for a 1D sensor.
- Because each “frame” is only one column wide, data rates are manageable; RAM-based buffering with later SSD dump is considered sufficient.
- One estimate: at 10,000 px height and 24-bit color, 40k lines/s is on the order of 10 Gbit/s.
- Discussion contrasts this specialized setup with general-purpose high‑speed cameras and old film-based strip cameras.
Rules, fairness, and what counts as winning
- Officially the torso decides the finish, not “any body part”; detailed guidelines define where the torso begins/ends relative to shoulders and hips.
- Some question whether camera angle or occluded body parts (e.g., hidden shoulder) could bias results, suggesting overhead or dual-side cameras.
- Others note that major competitions already require at least two finish cameras on opposite sides.
- There is debate over whether ultra-fine timing (e.g., 0.005 s differences) reflects true superiority or just noise and luck.
- Some argue for coarser official resolution or accepting more ties; others insist the race’s purpose is simply to identify who crossed first, however small the margin.
Alternative technologies and limits
- Suggestions include lasers, digital transponders on the chest, or dense rangefinder meshes.
- Pushback: transponder accuracy is limited (e.g., centimeter-level), body orientation/lean complicates “center of chest,” and sensors can’t yet match the optical precision and interpretability of photo finish.
- Several note that track construction tolerances, wind differences by lane, and reaction-time variation already impose a fairness limit beyond timing precision.
Critique of the article and broader context
- Multiple commenters find the linked article shallow and marketing-driven, lacking explanation of strip photography, optics, calibration, lighting, and reliability considerations.
- Others share links and anecdotes about older strip cameras in racing, industrial line-scan uses, drift-scan astronomy, and extremely high-speed DoD cameras, emphasizing that the underlying technique is mature but still fascinating.