Watch anything cut by somebody who does this for a living and you will see it: on the one sentence that matters, the frame gets slightly closer. Nothing else changes. It is the cheapest trick in editing and it is everywhere, which is why its absence is what makes a clip feel machine-made.
It is not about seeing the face better.
A talking head is a still picture with a mouth in it. Thirty seconds of that is thirty seconds in which nothing on screen changes, and a viewer who is half deciding whether to keep watching has nothing to hold on to. The push-in is a change. It arrives, it means something is happening, and it does that before a single word has been understood.
It also tells the viewer where to look, which is the part people miss. The frame moving in on somebody is the visual equivalent of leaning forward. It says: this bit. Cut the same sentence without it and the sentence is still there, but nothing marked it as the one.
Emphasis is a contrast. Two of them in one clip is not twice the emphasis, it is none.
The failure is easy to spot once you know it: a clip that zooms every few seconds. It reads as nervous rather than as deliberate, and by the third one the viewer has stopped noticing the move at all. The push-in works because the rest of the clip is still. If the rest of the clip is not still, there is nothing for it to work against.
The second rule is size. A push of a few percent reads as intent; a push of thirty percent reads as a mistake, or as an attempt to hide a bad crop. The move should be something a viewer feels and could not describe afterwards.
And the third: not every clip has a line worth it. A calm explanation with no turn in it is better left alone. Adding emphasis to a sentence that does not carry any is how a clip ends up looking like it was edited by somebody who was not listening.
One line per clip, or none at all.
Every clip is read for the one phrase an editor would push in on: a punchline, a number, the moment an argument turns. Where there is one, the frame comes in over about four frames, goes a fraction past where it is going and settles back, holds for the length of the line, and is gone again in a single frame. It aims at the speaker's head rather than at the middle of the picture, because the middle of a vertical frame is somebody's chest. A restrained whoosh lands with it, low enough to sit under the voice rather than on top of it.
Where there is no such line, the clip is rendered exactly as it would have been, untouched. That half is the part worth insisting on: a clip that is simply a person explaining something calmly gets nothing, on purpose.
It is a setting, on the video and on any template, so a channel that does not want it never sees it. And it is a tick in the editor per clip, which is where you will want it the day the machine marks a line you would not have.
Worth knowing, whatever you cut with.
In any editor the move is the same: a keyframe on the scale of the clip before the line and a second one a few frames later at a slightly larger value. Two things are worth copying from people who do this for a living. Let the move go a little past its final value and settle back over a couple of frames - a pure ease is what a computer does, and overshooting is what anything with weight does. And do not animate the way back: either hold it until the next cut, or drop it in one frame. A push that glides back out on a timer is the single clearest sign that nobody chose it.
Check it at the size people will watch it, which is a phone. A push that looks timid on a laptop is usually right; one that looks right on a laptop is usually too much.
A small, quick push of the frame toward the subject, held for a moment and then released. It is used to mark the line that matters. In short-form it is nearly always a scale on the clip itself rather than a real camera move.
Not when there are more pixels in the recording than in the finished clip, which is the normal case: a vertical clip keeps a slice of a much wider frame, so a few percent closer is still drawing on detail that was always there.
Less than feels like enough while you are editing. Something like a sixth of the frame, arriving over three to five frames, is where a viewer registers the change without being able to name it. Anything a viewer can describe afterwards was too much.
Usually yes, and quietly. A fast move with every frame perfectly sharp and nothing on the soundtrack is the thing that reads as a glitch rather than as an edit; a short whoosh under the voice is what makes it deliberate. Loud enough to notice is too loud.
Because nothing in it asked for one. The push-in is only added where there is a clear line to put it on, and a clip that is a calm explanation is better without. That is a decision rather than a gap.
Yes, on the video, on a template so every clip from it inherits the choice, and per clip in the editor.
The opening is a decision about which second to begin on, and it matters more than anything you can add on top of it.
Where the frame sits in the first place, and what happens to a wide recording that has to become a tall one.
The other thing carrying the clip while the sound is off.