Talking Head With Jump Cuts
A tight talking-head video with the silences cut, punch-ins and captions.
- Pipeline
- Basic edit (cut)
- Format
- 9:16
- Upload
- One or more takes of you talking to the camera.
The prompt
Type your own words into the fields: the [BLANKS] fill in as you go. Copy it and send it to Claude or ChatGPT with Cut Express connected.
Prompt
I recorded a talking-head video about [TOPIC]. Inspect it with a transcript and use the speech cuts. - Keep only the best take of each sentence and cut every silence and "um". The finished video is about [LENGTH IN SECONDS] seconds. - Start on "[HOOK LINE]" even if I said it later. - A punch-in zoom (about 1.2x) every 2 or 3 cuts, on the words that matter, so it never looks like one static shot. - Captions: bold preset, upper case, 2 to 3 words per line, highlight [CAPTION COLOR], just under the middle of the frame. - Clean my voice: denoise 40 and enhance. [LOOK] look, a little sharpening. Show me the timeline (each kept line with its time) before you render.
The blanks
- [TOPIC]
- What the video is about, in one sentence.
- [LENGTH IN SECONDS]
- How long the finished video runs, as a number of seconds. Reels and Shorts hold best between 15 and 60.
- [HOOK LINE]
- The first words on screen, the reason to keep watching, in under ten words.
- [CAPTION COLOR]
- The color of the highlighted word in the captions, as a name or a hex code like #FFD23F.
- [LOOK]
- The color look by mood, for example "warm and soft", "teal and orange", "crisp and bright", "black and white", "film".
Tips
- Say which take you prefer when you repeat a line ("the last take of each sentence"), or your AI will choose the cleanest one.
- Punch-ins on a word work better than a constant slow zoom: they mark the beat of the sentence.
- If a word gets cut, say "keep the end of the sentence at 0:42" and the next version fixes only that.
New to prompting video? How to write a good prompt for Cut Express covers the story, the hook, captions, sound, pace and how to iterate with versions.