How Coaches Break Down a Play Frame by Frame

Two different jobs, two different setups

Coaching analysis splits into two tasks that need opposite settings, and mixing them up is why the process often feels clumsy.

Technique work is about one athlete and one movement: the position of the trailing arm at release, whether the plant foot is under the hip, where the head is at contact. You need a handful of consecutive frames at high temporal resolution — the interesting window is often under half a second.

Tactical work is about where everyone was and when: the shape at the turnover, whether the far-side runner was in position three seconds before the pass. You need frames spread across many seconds, and one image per second is usually plenty.

The first calls for frame stepping. The second calls for interval extraction. Setting up for one and trying to do the other is the source of most of the frustration.

Setting up for technique work

Get the frame rate right first

This is the setting everything else depends on, and it is the one most often wrong.

In the frame-by-frame tool, Source FPS tells the stepping controls how far one step should move. It defaults to 30. If your phone recorded at 60, or the camera at 50, or you shot slow motion at 120 and it plays back at 30, stepping will not correspond to real frames.

Two symptoms tell you it is wrong. If the timestamp advances but the picture does not change, the value is set too high and you are stepping in fractions of a frame. If motion jumps noticeably between steps, it is set too low and you are skipping frames — which means the frame you are looking for may be one you never see.

To find the true rate: most phones record 30 or 60 by default, 240 in slow-motion modes. If you are unsure, open the file in VLC and check Tools → Codec Information, or step ten times in the tool and watch the timestamp — ten steps totalling 0.333s means 30 fps, 0.167s means 60.

Why 30 fps is often too coarse

At 30 fps each frame is 33 milliseconds. Plenty of decisive events are shorter than that. Bat-on-ball contact, the instant a foot plants, the frame where a hand first touches a ball — these can fall entirely between two frames, and no tool recovers what the sensor did not record.

If technique analysis is a regular part of your work, record at 60 fps as a baseline and 120 or 240 for contact-moment work. The trade is resolution and light: many cameras drop to 1080p at 120 fps and need considerably more light. For most field work, 60 fps at 1080p is a better default than 30 fps at 4K.

The stepping routine

1. Seek roughly. Use the scrubber to get within a second of the moment.
2. Step past it deliberately. Go a few frames beyond, then come back. Approaching from both directions makes it much easier to identify the actual first frame of contact rather than the first one you happened to notice.
3. Note the timestamp. The millisecond readout is what lets you find the same moment again next week, and what lets you compare the same athlete across sessions.
4. Save the key frames. Typically three: the frame before the event, the event itself, and the frame after. That triplet communicates far more than the single frame does.

Setting up for tactical work

Here you want a contact sheet: the whole sequence laid out as stills you can scan at once.

1. Load the clip in the frame extractor.
2. Set the time range to the passage of play — typically 10 to 30 seconds.
3. Set Time Interval to 1 second for general shape, or 0.5 for a fast transition.
4. Check the frame count. Twenty seconds at 1-second intervals is 20 images, which fits on one screen. That is the target — a contact sheet you cannot see at once is not doing its job.
5. Extract, then browse the preview grid. Structural problems — a gap that opens three seconds before the goal, a player who stops moving — are visible in the sequence in a way they are not in playback.

The reason this works better than watching the clip repeatedly is that stills remove time pressure. In playback your attention follows the ball. On a contact sheet you can look at the same off-ball player across twenty images in a row.

Turning frames into feedback

Frames on their own do not change behaviour. What tends to work:

• Pairs, not singles. A frame of the athlete next to a frame of the same moment done well — by them on a better rep, or by a reference athlete. One image invites argument; two invite comparison.
• One frame, one point. If you are showing a still to make three points, you need three stills. Attention on a single image is short.
• Annotate lightly. A line along the shoulders, a mark where the foot landed. Heavy annotation gets debated; a single line gets seen.
• Include the timestamp. It lets the athlete find the moment in the full video themselves, which is where the learning consolidates.
• Keep a per-athlete folder. The same technique frame across a season is the most convincing progress evidence there is, and it costs nothing but consistent naming.

Practical notes from the field

• Fix the camera position between sessions. Frame comparisons across weeks only work if the angle is comparable. A marked tripod spot is worth more than a better camera.
• Record longer than you think. Starting the recording late is the most common way to lose the frame that mattered.
• Trim before extracting. A two-hour session at one frame per second is 7,200 images. Set the range to the passage you care about first.
• Extract at source resolution. You will want to crop into a still to show a hand or foot position, and crops from 4K survive it.
• Work in chunks on long sessions. Frames are held in memory until you download them; take ten minutes at a time rather than the whole recording.

Footage of minors, and footage of other people

Worth stating explicitly because it applies to a lot of coaching video.

Recordings of young athletes are personal data and usually sit under a club or school policy about who may hold them and where they may be stored. Uploading that footage to an online converter is often a straightforward breach of such a policy, even when nobody intended any harm.

Browser-based extraction avoids the issue by construction: the file is read from disk into the page and never transmitted, so no third party receives it. You can verify this yourself — load the page, disconnect from the network, and extract; if it still works, nothing was leaving the machine. The verification steps take about a minute and are worth doing once.

Separately, footage of opposition players, broadcast material, and anything recorded at a venue with its own filming rules carries its own permissions. The tools here process files you already hold; they do not grant any right to record, keep, or share the content. Check the club policy before the analysis workflow, not after.