Quick Cut: The Case Appears
A twelve-second case film holds maybe twenty shots, and everything depends on whether the cuts sit on the beat. Start with tempo: pick something between 120 and 140 BPM, which works out to one beat every 0.43 to 0.50 seconds, so a 12.4-second timeline carries roughly 25 to 29 beats. Sixteen to 24 cuts go on the strong ones and the weak ones stay empty. Shots run 6 to 14 frames, every cut lands within a frame, and the cover frame comes from the 0.4 to 0.8 second window. None of these numbers is discovered afterwards; all of them are decided before the first clip is dragged in.
01Tempo first: turn 120–140 BPM into a beat ruler
Before a single clip goes on the timeline, build a ruler out of the track. At 120 BPM a beat lasts half a second; push to 140 and it shortens to 0.43. Tapping four beats with your foot against the low end beats auto-detect in most editors, especially when the drums are syncopated. Once the pulse settles, drop a marker on every beat — a 12.4-second piece usually ends up with 25 to 29 of them.
Those markers are candidates, and the count is bounded at both ends. Sixteen cuts is the floor: fewer and the case does not appear often enough to register as a range. Twenty-four is the ceiling: past that each shot drops under 0.3 seconds and the eye stops resolving shape. That works out to 0.6 to 0.9 cuts per beat, which is another way of saying hit the downbeat and let the rest breathe.
02A frame early or a frame late: 40 milliseconds decides
Picture arriving before sound reads as a shove from the drum; sound arriving first reads as drag. At 25 fps a frame is 40 milliseconds, so shifting the whole cut pass one frame ahead of the transient tightens the impact, while one frame behind loosens the same shots. Place markers on the waveform peak, then nudge the cut layer forward once — as a batch, never shot by shot.
That global nudge leaves half a beat of dead air at the tail, so build the edit backwards instead. Fix which beat the closing shot lands on, then walk forward in 6 to 14 frame steps. Any error collects in the first half second, before viewers start judging. Build front to back and the error ends up at the close, where the last case gets half a frame of face time.
03Long, short, short, long: spending 6 to 14 frames
Six frames is 0.24 seconds, enough for a silhouette or a latch opening. Fourteen is 0.56, enough for a front face plus a handle. Split the film into four or five groups and shape each one long, short, short, long: a wide to establish, two brief inserts on hardware and foam, then a wide to close. The eye needs that last resting place or it chases every cut.
Consecutive shots should differ by at least two shot sizes. Wide into medium reads as a new vantage point; wide into wide reads as the same box jittering. When the material will not stretch that far, insert a six-frame close-up rather than a transition effect. Effects eat the 40 milliseconds you were protecting and add a render pass.
04Waveform finds the beat, frame numbers hold the cut
Do not trust your ears for alignment. Zoom the audio until individual drum hits are visible, read the peak column as the transient, and cut on that frame boundary — error is zero. One frame of drift is survivable because 40 milliseconds vanishes inside a busy groove, but at two frames the picture smears. Check by stepping through with frame numbers on screen and the sound off, hunting for the one cut that jumps.
Source frame rate is the other half. Shoot at 50 fps, finish at 25, and each output frame maps to two source frames, so cuts must land on even source frames or you get a half-frame blend. Conforming footage to 25 fps before cutting beats cutting at 50 and dropping frames afterwards, which turns a one-frame error into visible stutter.
05Cover frame between 0.4 and 0.8 seconds, export at 12–20 Mbps
The first frame is almost always the wrong cover. It is typically empty or mid-blur; take the one sitting between 0.4 and 0.8 seconds, when the case is fully in frame and the latch has not moved. Slice five stills at 0.1 second intervals and compare them at phone-list width. Whichever keeps the silhouette readable wins, and it does not have to be the opening frame — upload it separately where the platform allows.
Export between 12 and 20 Mbps. Under 12, high-frequency detail in the fast cuts turns to blocks and every latch edge smears; over 20, the platform's second pass erases the difference and you have only added weight. Play the file back once and inspect three frames either side of each cut. Ghosting or banding means a re-export, not a re-upload.
| Tempo | 120–140 BPM (0.43–0.50 s per beat) |
|---|---|
| Shot length | 6–14 frames (0.24–0.56 s) |
| Cut count | 16–24 over 12.4 s |
| Cut tolerance | within 1 frame (about 0.04 s) |
| Shot-size jump | 2 steps or more |
| Cover frame | 0.4–0.8 s |
| Export bitrate | 12–20 Mbps |
| Beat count | 25–29 beats |
FAQ
What if the track has no clear kick?
Follow the bass note attacks or tap a four-beat loop by hand, then compare against the editor grid. Skip any bar that drifts more than half a beat, take the steadiest 8 beats as reference, and extrapolate at 0.43–0.50 second intervals without changing that reference midway.
Is 12.4 seconds mandatory?
It is an example, the ratio is what matters. At 120–140 BPM a clip of that length only holds 16–24 cuts, so if the runtime doubles the cut density has to come down. Otherwise shots compress under 6 frames and viewers see flicker rather than a case.
Will fast cuts make the case unreadable?
Unreadable comes from flat shot sizes, not speed. Jump at least two steps between neighbours, let each short shot carry one piece of hardware, and let the wide shots hold the outline. A six-frame close-up between groups gives the eye somewhere to rest.
Can this table be used on a 30 fps project?
Yes, convert everything to frames. At 30 fps a frame is 33 milliseconds, so one frame of error is tighter; keep shots at 0.24–0.56 seconds, which becomes 7–17 frames. The beat map stays identical and only the frame counts move.