See where the race was won.

RaceDev lines up two riders' POV runs frame by frame, with the live gap and both race lines drawn on the trail in front of them. Use it for highlight replays while the event is running, or a deeper breakdown once it's over.

Get in touch

r/r/MTB · Reddit

Race line comparison from GoPro footage#1 post of the day

21825

iillepic

I could watch hours of this. Please post more.

SSudeb-Roy

May I download/pay for the Beta version of this right now?

SScott0814

Looks sick so far, keep up the good work!

218 upvotes · 25 commentsView on Reddit

The response

People can't stop watching

I built RaceDev as a training tool for racers to analyze line choice. Then I started showing it around and people were captivated, and I realised the bigger draw could be media.

Watching the projected line get taken puts you inside a rider's head. You see the preparation turn into execution.

Both runs are time-warped so the riders are riding the same sections simultaneously. The clock tells you how it's playing out, not the videos drifting out of sync.

The product

What you can see

A rider's POV at Whistler with their race line, the other rider's dashed line, the gap and speed

The method

How it works

Raw footage next to the detected tire and its contact point, slowed down — What the camera seesWhat RaceDev sees

Contact pointWhere the tire meets the ground

The front tire is found in every frame, along with the point where it touches the ground. Each race line starts there and is projected into future frames.

Both riders' frames with the same landmarks marked in matching colours — Rider ARider B

Run alignmentMatch the frames

Landmarks on the trail are matched between the two riders' cameras, and each colour marks the same spot in both. That lines up every frame for both riders for easy comparison.

Raw footage next to the detected bars, hands and tire — What the camera seesWhat RaceDev sees

Rider detectionProcess what matters

The cockpit and tire are picked out of every frame, so the trail can be analyzed completely independently for accurate line placement.

Raw footage next to the race line ending at a jump's lip and returning on landing — What the camera seesWhat RaceDev sees

Airborne detectionAirtime

Every jump is detected and timed. This stops the line from showing where it shouldn't and is another metric for comparison.

The analysis

It calls the race too

The race data from every pair is analyzed and flags the key sections where the race was decided. A commentator or editor can go straight to the moments that matter.

Accuracy target

0.01s

Target gap accuracy on 50fps+ source footage, validated against official split times.

Working together

What we can create

For an event

From your original camera files.

Where it was wonThe podium runs side by side, ready after the final.
Live broadcast replaysRider against rider clips between runs.
Course previewA practice run so viewers know the track before racing starts.

For media and creators

From two POV runs on the same trail.

Run breakdownsThe gap, both lines and the key sections, with the numbers behind them.
Section reelsThe sections where the most time moved, ready to post.
Pro breakdownsA pro talks through a race run line by line, for more in-depth analysis.

Who's building it

Tod SleemanFounder, RaceDev.ai

I'm a software engineer living in Squamish, BC. I ride and race downhill/enduro, and trail build as a volunteer.

Originally from England. I grew up reading MBUK religiously and was building "North Shore" style trails before I even knew where the North Shore was. I'm still doing what I was doing at 12, riding every day, along with building the tech to take mountain biking further.

LinkedIn Instagram

Let's show the line

I'm looking to partner with event organizers, broadcasters, and media teams who want to give viewers a deeper view of the race.

Discuss a pilot