Video Feedback: Filming One Question at a Time

Filming yourself rowing looks simple and usually is not. The clip you end up with either shows too much to settle anything, or stops before the thing you wanted to see has happened. The camera is not the problem. The question is.

The context

Your erg already tells you a great deal. The monitor's all-data screen already carries your pace, time, distance, average pace and stroke rate, and its force-curve display plots the power you apply across the stroke, which the manufacturer describes as a helpful tool for improving your technique ([3] Concept2 Understanding the PM5 2024). A video clip adds none of that. What it adds is the part the numbers cannot carry: shape, position and sequence, the things a coach can see and a number cannot.

That is worth doing carefully, because the picture and the numbers are easy to confuse. On-water rowing biomechanics has enough variables in play that the research literature has had to organise itself: a scoping review of the field sorted everything reported into time, space and force, and found gate force, horizontal oar angle and boat velocity to be the most frequently measured ([6] Legge 2024). A phone on a shelf in your hallway is a fraction of that, and it cannot answer a question it was never framed for.

The manufacturer's own technique material splits the stroke into a catch, a drive, a finish and a recovery, and describes the drive as the work portion and the recovery as the rest portion that prepares you for the next drive ([1] Concept2 Rowing Technique 2024). That division is the natural unit of a question. "Is my lower back rounding at the catch?" can be answered by a clip. "Is my rowing good?" cannot.

The core idea

One clip, one question. The clip is not a performance to be judged. It is a measurement of a single thing, and every part of the setup — where the camera stands, what the frame contains, what rate you row at — is a decision about what that measurement will be able to show.

Start with the camera, because that is the part people get wrong. Two-dimensional video has been validated against three-dimensional motion capture in a sport with a similar problem: the 2D knee angles in the sagittal plane correlated almost perfectly with the 3D values, at r = 0.93 to 0.99, but the ankle angle carried a large bias and the rear hip angle correlated only moderately, at r = 0.31 ([7] Chida 2024). A camera that is not square to the plane of the movement will hand you numbers that look like measurements and are not. Camera position and height are not a detail. They are the measurement.

So choose the view from the question. A question about the shape of the stroke — the catch, the finish, how the trunk arrives at the end of the drive — wants a camera square to the side of the rail, level with the seat. A question about symmetry — one knee higher than the other, a handle not level, a pull that favours one side — wants a view down the line of the machine, from the front or from behind.

Then choose the rate, because the shape of the stroke is not the same at every rate, and the part you want to watch is short. The force in a rowing stroke is produced over roughly 0.3 to 0.4 seconds ([4] Secher 1993), so most of what you want to see is over before the camera has registered it. In elite rowers working on two different ergometer setups, the leg catch angle and the trunk finish angle both changed with stroke rate, with significant interactions between rowing modality and rate for both ([5] Penichet-Tomas 2025). A clip filmed at a rate you would never train at is a clip of a different movement. If the question is about shape, drop the rate until the shape is legible. If the question is about whether you can hold a position at a training pace, film at that pace.

The review is the third decision, and it is the one most people get wrong by default. Two findings pull in the same direction. A conceptual analysis of the feedback literature concluded that the type of feedback and the way it is given can be differentially effective ([12] Hattie and Timperley 2007). And in a meta-analysis of feedback interventions, more than a third of them made performance worse, with effectiveness falling as attention moved from the task towards the self ([11] Kluger and DeNisi 1996). Both of those point the same way. Ask about the stroke, not about the rower.

The how-to

1. Write the question down before you touch the camera. The manufacturer breaks its common technique errors into three groups — arms and grip, back, and legs ([2] Concept2 Improve Your Rowing Technique 2024). That list is a good menu precisely because each line is already a question about one part of the stroke. Take one line. Write it on a piece of paper.

2. Set the machine up first. Foot stretcher, seat, damper, monitor, all of it exactly as you would have it for a normal session, because a clip taken on a machine set up for filming is a clip of a different rower. Set it, then look at the monitor and confirm the pace is the one you expect before you record anything.

3. Put the camera square to the rail. Side-on for anything about shape, at seat height, far enough back to take in the whole stroke and close enough that the handle and your feet are big enough to read. Whatever you are filming, check one thing before you record: is the camera square to the rail, or slightly off it?

4. Film a run of strokes, not one stroke. A single stroke is a single sample, and the stroke you want to know about is the one you make when you are not thinking about it. Record enough strokes to see whether the thing repeats, and note the rate in the clip's name so you know what you are looking at a week later.

5. Review once, after the set, against one criterion. Play it back slowly, look for the one thing you wrote down, and write a sentence about what you saw before you watch it again. The case for choosing when to watch is a real one: in a study of the basketball set shot, learners who could request their own video feedback of their form scored significantly higher on form in the transfer phase than a group whose feedback was scheduled for them ([8] Aiken 2012). What you should not do is watch every rep as you go. In experiments on motor learning, feedback delivered during practice interfered with learning by creating a dependency on the visual feedback itself, while the groups that received feedback at the end of practice, or chose when to receive it, made significantly smaller errors on a retention test a week later ([9] Akizuki 2025).

6. Change one thing, then go and row it again. Not a list of things. One change, from one clip, for the next few pieces. You will learn far more from one question answered three times than from three questions answered once.

The trade-offs

Two limits are worth naming, and one of them is a warning about the method rather than the equipment.

The first is that choosing when to look at your own footage is not automatically the better method. The support for letting learners control their own feedback schedule comes mostly from laboratory comparisons whose control groups do not look much like a real coaching session, and when a more applied comparison was run — sixty participants, three groups, delayed post-tests — no group differences appeared in either the radial error or the bivariate variable error ([10] Yantha 2022). The practical reading is that choosing when to watch is a reasonable default, not a law. The thing that matters is that the clip answers a question.

The second is that a home clip is an illustration of your stroke, not a measurement of it. The validity figures above come from a camera set up and calibrated for a specific question in a laboratory. In a hallway, at whatever height your shelf happens to be, the clip is good enough to see shape and not good enough to measure angles. Use it to know what to change, and use the monitor for the numbers. They answer different questions and they are both more useful when they are not asked to do each other's job.

And keep the footage where you want it. Film in your own space, on your own device, and delete the clips you do not need. Our official Privacy Policy is the reference for anything you choose to share with us, and it is a better answer than any summary of it.

What to do tomorrow

Take one line from the manufacturer's list of common technique errors. Set the erg up the way you train on it. Put the phone square to the rail at seat height, far enough back to see the whole stroke. Row a set at a rate you would actually train at. Then watch it once, for that one line, and write one sentence about what you saw.

Tomorrow, take the next line and do it again from the same camera position, so the two clips can be compared. If the first sentence was "my back rounds before the catch" and the second was "my back rounds more when the rate is up", you have something to act on. If both sentences are about your back, you have a fault and a question, which is where this is supposed to end up.