Camera-follow captions

Each sentence written across a canvas far bigger than the frame while a camera follows the words as they are said: every word a step bigger than the last with the camera pulling back to keep it the same size (or the camera gliding along lines of text), motion blur on every move, and a pull-back at the sentence's end to show it whole. Give a transcript with --words; *starred* words and the ones in `accents` take the accent.

captions/camera-follow
Captions
captions
Overlay with alpha
6 s (1.5–120 s)

Camera-follow captions at 16:9 — Playing live with volito/browser, the engine a render uses. Each tab lays the item out again for that shape, not a crop. The graphic is transparent; the grey stands in for your footage.

Render it

npx volito render captions/camera-follow --words talk.json --out camera-follow.mov

Captions take their timing from a transcript: --words reads Whisper JSON or [{w, s, e}], one entry a word. Add text to caption your own wording on the voice's timing; without a transcript, text runs at pace seconds a word. Starred words and the ones listed in accents are the key words.

That writes ProRes 4444 with alpha, ready to drop over footage. The same item in another shape, brand or format:

volito render captions/camera-follow --words talk.json --aspect 9:16 --style cobalt --out camera-follow-9x16.mov
volito render captions/camera-follow --words talk.json --format mp4 --out camera-follow.mp4
volito still captions/camera-follow --t 4 --out camera-follow.png
volito info captions/camera-follow     # every prop and part
volito add captions/camera-follow      # copy its source into ./motion to edit

From Node:

import { findItem, readWords, renderVideo } from "volito";

await renderVideo(findItem("captions/camera-follow"), { words: readWords("talk.json"), out: "camera-follow.mov" });

Or ask an agent with the MCP server connected: “Render the Volito camera-follow captions (captions/camera-follow) in 9:16, in our brand.”

Props

Set each with --prop name=value, or all at once with --props and a JSON object.

Prop	Type	Default	Takes
`text` what is said; leave empty to caption the transcript as it is. Wrap a word in stars for the accent	longtext	Captions that follow every word you say, right on time.	≤ 600 chars
`accents` Accent words words to put in the accent wherever they are said, comma separated	text	—	≤ 120 chars
`camera` grow: each word a step bigger, the camera pulling back to keep up · track: lines of one size, the camera gliding along them	choice	grow	grow \| track
`framing` how big the newest word is on screen	choice	standard	standard \| tight \| wide
`growth` grow camera: how much bigger each word is than the last	choice	standard	standard \| steady \| bold
`blur` Motion blur	choice	standard	standard \| subtle \| heavy \| off
`ending` Sentence end pull back: the camera shows each sentence whole once it is said · hold: it stays on the last word	choice	pull back	pull back \| hold
`ground` Behind the words vignette: a soft dark edge over the footage · none: straight on the footage with a shadow · panel: the brand's ground, full frame	choice	vignette	vignette \| none \| panel
`case`	choice	upper	upper \| as typed \| lower \| style
`pace` Seconds per word without a transcript	number	0.42	0.1 – 2
`ticks` Sound on every word	boolean	false	—

Parts

Each part takes any move, ease or idle: --motion ground.in=pop.

Part	What it is
`ground`	the vignette or panel behind the words
`words`	the words and the camera over them

Tags: captions, kinetic typography, camera, zoom, transcript, words, social, reels, hook

More in Captions

Karaoke captions

Social captions on the voice: a line or two at a time, the word being said lit by a pill that slides from word to word (or in the accent colour, or underlined), the rest of the line waiting dimmed. Give a transcript with --words; *starred* words and the ones in `accents` take the accent.

Pop captions

Big captions a few words at a time, each word landing on its own time with a pop — the newest one in the accent. Give a transcript with --words; *starred* words and the ones in `accents` take the second accent.

Subtitles

Clean subtitles timed to the voice: up to two lines at a time in the brand's text face, broken at pauses and sentence ends, on a dark box behind each line or with a shadow. Give a transcript with --words.

Marker captions

A highlighter swipes under each word as it is said and stays lit, a little crooked like a real pen, so the line fills with the brand's colour as it is spoken (or an underline stroke). Give a transcript with --words; *starred* words take the second accent.

Neon captions

The line hangs as unlit glass tubes and each word flickers on as it is said, glowing in the accent (accent words in the second accent) and drifting a hair once lit. Give a transcript with --words.

Decode captions

Each word arrives as scrambled characters that settle into the word letter by letter as it is said, in the mono face (or the display face). Give a transcript with --words.