01Eight to twenty clips covering four to eight items
Scrub to the end and count the cuts before anything else. 8 to 20 clips is the usual range, mapped onto 4 to 8 test items, so one item can be split across two or three clips joined by a transition. Clip count and item count seldom match, and that gap is the first evidence of choices: items that filmed well got two clips, the rest got one landing.
Clips run 5 to 40 s each. The short ones keep only the touchdown; a 40 s clip carries setup, placement, and the lid closing. Ask of every clip whether it shows process or outcome. One with setup lets you see how height was measured. One that starts mid-air leaves height to the caption. That single difference decides how much of the clip you can use.
02One to three units per item cannot carry a verdict
Any clip may stand for 1 to 3 units. That is enough to show what a single impact looks like, not enough to support a claim about a batch. With one unit you are watching an event. With two or three you start to see repetition, though you are still counting on one hand and the spread stays unknown.
Count units by markings, not captions. The same case carries the same scratch or sticker from clip to clip, so a spotless case in every shot usually means one unit per shot. Ten clean shots at 40 s apiece can come from a single case. A small sample does not make the footage dishonest; it caps how far the conclusion travels.
03Three to six digits give the clip a way back to its log
Cases usually carry a 3 to 6 digit number on a side or end panel, and now and then it drifts into frame for half a second. That half second is the only hard link between a clip and its record. Longer numbers narrow the search; 3 or 4 digit ones repeat inside a batch and need a date to separate them.
Copy down every number that appears and match it against the log. That confirms the clip is the run it claims to be. Where no number shows, ordering by sequence is all you have, and one wrong step shifts everything after it, sometimes across 2 h of footage. Label visibility beats image sharpness here. Half a second is enough, but only if it is legible.
04Half an hour to three hours, cut down to seconds
A single test — setup, calibration, release, teardown — often occupies 0.5 to 3.0 h of record. What airs is 5 to 40 s, a compression of tens to hundreds of times. Most of what disappears is preparation, waiting, and repeats, though an anomaly in the middle can vanish too, depending on what the editor kept. The ratio matters more than the raw length.
The timeline is what you can still reconcile. The log has start and end stamps; the cut has transitions. Line them up and the skipped stretches surface. Skipping is not the problem. The problem is that the remaining shots still run continuously, so the sequence reads as one take — and the distance between a 40 s clip and a 3.0 h log goes unnoticed.
05Speeding up to four times erases the window that matters
Speed is the compilation's main tool. Below roughly 1.5 times normal, action still reads. Past 2 times, the milliseconds around touchdown collapse into a smear: deformation, rebound, and rotation arrive too fast to separate. Treat any accelerated clip as an overview and never as evidence — an overview says what was done, not how well. Anything above that is a summary, not a measurement.
Retakes deserve a note too, 0 to 3 being common. Shooting an item twice or three times and airing the best is reasonable, but that clip then represents the best run, not the only one. If a retake count is stated, read the clip that way. If it is not, log the clip as a single unit and carry that number forward.
| Clips included | 8–20 clips |
|---|---|
| Single clip duration | 5–40 s |
| Units per item | 1–3 units |
| Test items | 4–8 items |
| Serial number digits | 3–6 digits |
| Raw log duration | 0.5–3.0 h |
| Retakes | 0–3 retakes |
| Playback speed | 1–4 times |
FAQ
If one clip passes, does the item pass?
No. A clip usually stands for one to three units, and single digits only show what that impact looked like. Whether an item holds up depends on its full record and the unit list, not on where the clip sits in the running order.
Why bother with the serial number?
It is the only hard link between footage and record. Three to six digits on screen for half a second is enough to tie the clip to one specific run. Without it you are guessing by order, and one bad guess spoils the rest.
Does playback speed change what I can judge?
It does. Past two times, contact and rebound compress until the shape is unreadable and the order of deformation and rotation is lost. Take accelerated clips as overviews, and find the normal-speed or slow-motion version when the motion itself is the question.
How should a retaken clip be counted?
Zero to three retakes is normal practice, but what airs is the chosen take. It shows the item at its best, not the spread of results. Record it as one observation in your tally rather than several.
