Video editing
Talking to the camera: shorts that hold attention
One face and one claim is the cheapest short-form video there is — and the format most people sabotage. The retention is in the first line, the eyeline, and cutting the setup rather than the substance.
The first line is ninety percent of the video
A scroll decides in about a second, and what it reads is the first sentence, spoken, with the face already mid-thought. Which means the video is built backwards: write the payoff line first, then write whatever comes before it only if the payoff needs it. "Here's the thing nobody tells you about X" is a first line. "So, um, today I wanted to quickly talk about…" is a swipe.
Practically: record the claim as its own take, and put it as the clip's first frame. Most talking-head shorts are a sixty-second warm-up followed by a good sentence; the edit is just promoting that sentence and deleting the warm-up — the cut-the-setup principle from the the Podcast clip guide, applied before you record instead of after.
The lens, not the face
Eye contact is the format's superpower and almost nobody aims it right. The rule is unforgiving: look at the lens, not at yourself on the screen. The difference reads as the difference between someone talking to you and someone talking near you. Phone cameras make this worse — the lens is centimetres from the selfie screen your eyes keep drifting to. Tape a dot next to the lens if you have to; it works within a day.
Energy runs at about eighty percent of "on stage" and about one hundred and fifty percent of "normal conversation" — the feed is intimate but the medium is noisy, and the register that survives is a friend telling you something they're excited about. The single biggest energy-killer is reading. If you need a script, record it in sentences, not paragraphs: one breath per take, and the jump cuts hide the seams better than any prompter hides the reading.
Cut once per breath, not per second
The fast-cut style — a zoom or a whip every two seconds — is an arms-race relic; the talking-head comeback in the short-form trends post killed it. What replaced it is rhythm: cut the dead air between sentences, cut the false starts, cut the setup of each point, and then let the face talk. A jump cut at every natural breath keeps the pace without the stutter, and leaves the viewer's attention on the argument instead of the editing.
Silence inside a sentence is a different instrument: keep one pause before the payoff and cut every other. A pause at the right moment does what a zoom used to; three pauses in a row just reads as a video that needs an editor.
Frame it once and never think about it
Vertical talking-head framing has one setting worth remembering: eyes on the upper third, chest up, a little headroom, subject centred. That is the whole thing. Everything else — background, depth, lighting — should be boring enough that captions and the face do the work. A ring light off-axis by forty-five degrees, a window if it is free, face lit and background darker: that is the complete beginner lighting kit.
The reason centring matters more than beauty: if this recording is going to be sliced into shorts later — and a solo talking-head video is the most sliceable format there is — a centred subject means every 9:16 crop lands on the face without tracking at all. The crop arithmetic is in the reframing guide; shooting for it makes the problem vanish.
Captions carry the other half of your audience
For a solo video the caption is not an accommodation for muted viewers, it is a second performance: the highlighted word lands the emphasis the voice would have. Word-timed, two to three words per frame, high enough that the app's own buttons do not sit on top — the placement rules, again, in the the Podcast clip guide.
Typing them is optional, though. A solo clip of one claim, recorded in sentences, is the exact input a one-pass pipeline was built for: upload, get back the captioned, framed, quote-tagged version, post the same afternoon — which is what the free monthly video at Video To Reel exists for, and the reason the Talk to TikTok page exists at all.
Frequently asked questions
Where should I look when recording a talking-head short?
At the lens, not at your own image on the screen. Looking at the lens reads as eye contact to the viewer; looking at the selfie preview reads as distraction. A dot of tape beside the lens fixes the drift within a day of recording.
How fast should I cut a talking-head short?
About once per breath: cut the dead air between sentences, the false starts and every setup, then let the take run. Jump cuts that land on natural breaths keep the pace without reading as a stutter; cutting every two seconds is a style most feeds have moved past.
Do I need a script for a talking-head short?
You need a written first line; the rest can be bullet points you speak fresh. If you read a full script, record it one sentence per take — jump cuts hide sentence boundaries perfectly and reading badly from a prompter is visible in a second, while a cut between sentences is invisible.
Every how-to on this blog describes work you can hand to Video To Reel in one upload: the transcript, the moment picks, the 9:16 crop on the face in shot and the burned-in captions, returned as ready-to-post clips.