What I actually shoot
Hill climb and club rally, mostly in Denmark. A typical day is somewhere between five and twelve runs, and each run is short. A hill climb is over in under two minutes.
I run three cameras. Two GoPros and an Insta360 X5. One inside the car, the others outside on different mounts depending on what I'm trying to show that weekend. They're not synchronised in any way when they record. I start them one at a time, walking round the car, usually while someone is telling me the queue is moving.
So a day out produces somewhere north of thirty video files, from three different brands, none of which agree on what time it is.
The part that used to eat the evening
The filming was never the problem. The problem was afterwards.
Thirty-odd files land on the disk with names like GX010045.MP4 and VID_20251004_124128_00_023.mp4. Nothing in those names tells you which run they belong to. So the first job was always sorting: open a file, watch a few seconds, work out which run it was, drag it into a folder, repeat. Then line up the three angles for each run by hand, one at a time, twelve times over.
By the time any of that was finished I'd lost the evening and hadn't made a single video yet. That's the reason most of my footage from a couple of seasons still sits on a drive, unedited. Not because it was bad. Because sorting it was such a miserable prospect that I kept not starting.
That's really what StintBox came out of. I wanted the boring part gone.
What I do now
I copy everything off the cards into one folder and drop the whole lot in at once. All three cameras, all runs, unsorted.
The app works out which files belong together. It uses what the files themselves contain, meaning when they were recorded, where, what the telemetry says and what the audio sounds like, rather than filenames. That's the only approach that survives three different brands with three different naming schemes. What comes back is one group per run, with the three angles already in it.
Then it aligns each group. That's the step I care most about, because it's the one I used to do by hand twelve times a night. It handles the mixed brands without me telling it anything, and it flags the ones it isn't sure about instead of quietly getting them wrong.
I check the flagged ones. Usually that's one or two out of twelve, and checking means scrubbing to something I can see in two angles at once and stepping through a few frames. Then I export.
Five export formats per run: the full run, a vertical cut, and a few shorter framings for different places I post. On a twelve-run day the whole thing, from import through grouping, sync and check to export, is done inside an hour. Most of that hour is the export queue chewing through files while I do something else.
The bit I didn't expect
Getting the runs cut was the goal. The thing that actually changed how much I publish is what comes after.
Once a run is grouped and aligned, it stays that way. So when I want thirty seconds of a particular moment, like a corner that went badly, an overtake, or the one time the rear stepped out on the exit, I go back to that run, find the moment, and cut it. All three angles are already in step, and the telemetry overlay follows across all of them, so I can cut from the in-car view to the outside camera mid-corner and the speed readout doesn't jump or reset.
That takes a couple of minutes now. It used to mean re-doing the alignment for that clip, which meant it mostly didn't happen.
The practical effect is that a race day produces a handful of finished videos on the evening, and then keeps producing short clips for weeks afterwards whenever I want one. Same project, no re-work.
What still takes time, honestly
The export queue. Rendering three angles with overlays across twelve runs is real work for the machine, and no amount of software design makes that instant. I start it and go and do something else.
Checking the flagged syncs. It's quick, but it isn't zero, and I don't skip it. A run that's a few frames out looks fine in the preview and wrong in the finished video, and finding that out after exporting five formats of it is worse than spending twenty seconds up front.
And the actual creative decisions: which runs are worth posting, where to cut, what to leave out. That's the part I want to spend time on. It's the whole reason for getting the rest out of the way.
If you're running a similar setup
Two things I'd pass on, both learned the hard way.
Copy the original files off the card and work from those. Anything that's been trimmed in the camera's own app, exported from an editor, or re-encoded to save space usually arrives with the telemetry stripped out. It looks identical and it's much harder to work with.
And make one sharp, visible, audible event before you start, like a clap or a car door, with every camera already rolling. Then again at the end of the day if you can. It takes five seconds and it makes the awkward setups easy.
Beyond that: dump everything in one go and let the sorting happen on its own. The reason I built it that way is that I know exactly what the alternative costs, and it's a season of unedited footage.