Site logo - a 3d rendered tabletop jump

DirtScan

a GoPro - tiny but powerful

How a single GoPro can 3D scan a mountain bike trail

05/08/2026

By Matthew Hilton

No AI writing used

If you haven’t already, you should read the first post which explains what photogrammetry is.

One of the goals of DirtScan is to have a textured 3D model of a mountain bike trail which we can put on the web. At it’s core, we need a process which can produce this and:

  1. Produces the correct size/scale, matching the real world.
  2. Works outdoors.
  3. Is simple to do, commonly available, and cheap.

On constraint #3

Constraint #3 is somewhat of a self imposed constraint. One could argue that if you had enough money that this is a solved problem. Surveying, after all, is not a new field.

I imposed this constraint on myself as I believe it’s a critical piece in making DirtScan a useful project, as opposed to a niche one. It’s one thing for me to purchase some expensive equipment, but to ask people elsewhere to do this really limits potential contributions.

I’m imagining a time in the future where there could be people all over the world scanning and processing these trails and working together to create DirtScan.

Available options

  1. LIDAR (Light Detection and Ranging - uses lasers to get very accurate distance measurements)
    1. Commercial LIDAR scanners
    2. “Hobbyist” LIDAR scanners
    3. DIY LIDAR scanners
  2. Cameras (via photogrammetry):
    1. GoPro
    2. Phone camera
    3. Professional DSLR camera
    4. Drone camera
  3. iPhone (*Some iPhone pro models contain both cameras and a LIDAR sensor)

Out of these, almost all fail the constraint #3 - they are not easily available and not cheap. This excludes all LIDAR scanners and also excludes professional DSLR cameras - devices which in the future may be viable but are currently still at the $1k+ price range.

DIY LIDAR scanners are somewhat cheap, but they fail the ‘simple’ test - I can’t ask people with limited knowledge of tech to build their own LIDAR scanner.

I also exclude drones, which while could work and are actually quite cheap nowadays - in the context of national parks (where a lot of MTB trails are) drone usage is heavily regulated and controversial. Also, trees.

This leaves 2 main options, your phone, or a GoPro.

Out of these two, the GoPro wins for a very clear reason: it has a removable battery and removeable SD card.

For context, these trail scans take around 1hr/km to complete, and 1km of trail results in about 5,000 photos, about 15GB. Imagine if you needed to scan a 4km trail - this would require a lot of battery and a lot of storage (about 64GB).

Apple limits the amount of storage on a phone massively, and doesn’t allow any swappable batteries - meanwhile for a GoPro, this is what it was built for - a 512GB SD card is extremely cheap, and you can carry as many batteries as you want.

Maybe you could get away with a cheap Android phone with a SD card slot and swappable batteries, this might be an avenue for the future. However, GoPro actually wins slightly over even a cheap android phone, in that most mountain bikers would be more likely to have a GoPro already, rather than having to purchase a separate cheap android phone (constraint #3 - commonly available).

Interestingly, certain iPhone Pro models do include a LIDAR sensor. I did seriously look into this, but reportadly it is not great outdoors and has the same issues as above - not enough storage and not enough battery. Also, they are expensive!

What it is not great at

Cameras of all kinds fail (without anything else) in constraint #1 - They do not produce models that are true to scale. This is simply because the camera does not know how big something is. A LIDAR scanner on the other hand, knows exactly how big everything is (because of the way it uses light/lasers).

This can be fixed, however, by using what is called a scale bar - this is simply a “thing” that you include in your scans that is of a known scale. Using this, the computer can then correctly scale the model to the same as it is in the real world.

I’ll explain further the different types of scale bars and their pros/cons in a later article.

In summary

What is great about a GoPro for this task is:

  1. You probably already have one; older models can be picked up quite cheaply if not.
  2. It has removable SD cards and batteries
  3. It is very commonly available, no DIY of technical knowledge required.
  4. It has more data than a standard camera - it also includes GPS locations and IMU (gravity) data in every photo (I’ll explain why we can’t rely on this in a later article).

Given this, I’ve been using a GoPro for this project and don’t expect that to change any time soon.