Editing on the Phone: The Same Job in Two Apps — Beat Cuts, Real Slow Motion, Speed Ramps, Manual Colour, Transcript Editing and Export Settings That Survive Instagram

Smooth slow motion exists only where the source frame rate exceeds the timeline frame rate, and the ratio between them is the floor. 60fps on a 30fps timeline slows to 0.5x; 60fps on a 24fps timeline slows to 40% - 'so this is true slow motion'. Go below the ratio and the app has to invent or hold frames, which shows as gaps between them and reads as cheap. He calls over-slowing 'one of the biggest beginner mistakes that I see', and it is the same arithmetic taught on Day 1, now with a slider attached to it.

Songs have a rhythmic pattern, and cuts that land on it feel intentional while cuts that land near it feel sloppy. Both apps will analyse the track and drop markers on the downbeats; clip edges then snap magnetically to them, so aligning a montage becomes dragging rather than counting. He points out that this is exactly what the reels people admire are doing - 'they're adding the beat markers' - and demonstrates reading structure off the markers, counting four beats before a drop so the biggest shot lands on it.

The order is deliberate: assemble the clips, set the frame ratio and decide fit or fill, then reposition and re-crop each clip for composition - and delete anything that stops working once cropped, or that repeats a shot you already have. Only then do titles, colour and sound design get touched. The reason for doing the culling early is that everything after it is wasted effort on clips that should not survive. He closes the pass with one more run through the whole timeline to confirm the sequence before moving on, and reminds the student that the reel still needs 'a distinct beginning, middle and end'.

The basic slider changes speed across the whole clip. The curve or graph editor changes it at points - slow at the bottom of the curve, fast on the way back up - which is what a speed ramp actually is. 'Speed ramping is different from time remapping because we're actually not changing the speed across the entire clip. We're just ramping it either at one particular point or at multiple points.' The most common use is as a transition: let a clip return to normal speed and then accelerate into the next one, so the cut is carried by momentum rather than by an effect.

A LUT or filter tuned on one camera behaves unpredictably on another, so a preset that flatters drone footage will not flatter phone footage. His routine is manual: exposure first so the image is correct, then contrast and saturation until it 'pop[s] and... really kind of jump[s] off the page'. Then the more surgical controls - highlights and shadows, which 'become a lot more valuable tools than just overall exposure control' because you can lift a dark subject without losing the detail in the sky - and HSL when a single colour needs work rather than the whole image.

CapCut transcribes the dialogue and lets you delete pauses, flubs and restarts from the text, which removes the slowest part of cutting a talking head - hunting for the split points. He is honest about the limits: the automation 'sometimes leaves a little too long with breath marks', so a manual tightening pass still follows. Paired with it are two cheap devices for the same footage: jump cuts with a crop-in so the cut reads as intentional, and auto-captions, which are free and adequate even though many of the styles carry a Pro badge.

Without a green screen, the pattern is: duplicate the clip, promote one copy to an overlay layer and remove its background, keep the muted duplicate on the main track purely so the timeline keeps its timing and audio, and slide the new background art between the two. The rule that makes layers behave is stated plainly - 'remember, the lower the level, that's what actually shows through'. And the reason for the seemingly pointless duplicate: 'if I didn't have this duplicate here... the entire script would have moved up and the clips won't time up.' The pattern transfers to any layer-based editor.

Digital stabilisation counteracts motion by cropping in and warping frames, so every increase in smoothness costs resolution and risks the jello warp. The working rule is therefore to start at the minimal setting and only step up if the shot is still unusable - and to treat the feature as a rescue for a clip you really want rather than as a routine pass. It is the same conclusion the Day 5 movement lesson reaches from the other direction: shots too shaky to use should be reshot or dropped, not stabilised.

Instagram and TikTok re-compress what you upload, and they compress a large file harder. Exporting 1080p at the source frame rate therefore ends up looking better than exporting 4K, because you hand the platform something it does not have to fight. He targets under about 200MB for Instagram for the same reason. Frame rate matches the footage - 24 or 30 - and his own default is '1080p at 24 frames per second 99% of the time', with a bitrate anywhere from 10 to 30Mbps counting as high quality and 25 as his usual. HDR follows the source: on if you filmed HDR, otherwise leave whatever the app auto-detected.