Updated 2026-08-25

Why We Built a Dual Camera Recorder (And What It Cost Us)

Most people who want both angles buy a second camera. You already own the second camera — it is on the other side of the phone you are holding.

The idea was not clever. Point the phone at something, and the interesting reaction is happening on the other side of it. Every reaction video, every piece of vlog footage, every “you had to be there” moment is two shots that were never recorded together.

The part that is genuinely hard

iOS will happily give you the front camera or the rear camera. Asking for both at once means AVCaptureMultiCamSession, and that is a different animal from the standard capture session in three ways that matter.

It needs an A12 chip or newer, which quietly rules out a slice of devices still in daily use. It costs significantly more power and thermal headroom than single-camera capture, so a long session on a warm day is a real engineering constraint rather than a theoretical one. And it will not let you configure both inputs as freely as you can configure one — the system reserves the right to refuse a combination outright.

What that means in practice is that a lot of the work is not the feature. It is the negotiation around the feature: checking support before offering the mode, degrading cleanly when the hardware says no, and managing heat so a recording does not die at the interesting moment.

The decision that shaped everything else

Nothing leaves the device. No account, no upload, no sync.

That was not a marketing position first — it was a scope decision. The moment footage moves to a server you own a storage bill, a retention policy, a breach surface, and a support burden about someone else’s holiday video. For a two-camera recorder, none of that improves the product. It only makes it more expensive to run and more dangerous to get wrong.

The side effect is that the app works with the network off entirely, which turned out to matter more than expected once people started using it in cars and on dive trips.

The features we did not plan

Two of the modes people use most were not in the original sketch.

Dashcam mode came from watching how people actually mounted the phone. If it is on the dash recording the road, the front camera is pointed at the people in the car — so the reaction shot is free. Speed and GPS come off the device’s own sensors, which means it keeps working in a tunnel or out of coverage.

Black screen mode came from the opposite instinct. Sometimes the point is that nobody is watching a screen. The display goes dark, the recording continues, and the phone stops being a distraction to the room it is in.

What I would tell someone starting this

Build the capability check before you build the capture. We wrote the recording path first and the support matrix second, which meant every early tester on an older device hit a wall we should have anticipated.

And be honest in the store listing about the hardware floor. An app that cannot run on someone’s phone is a one-star review waiting to happen, and the fix is a sentence, not a sprint.