Place one yellow word per speech beat, low-centre, and keep the cadence running across hard cuts between shots

What the source does

Handheld event footage cuts between five shots (local ~28, 58, 70 and 93) while a single small yellow word sits at the lower centre of frame and changes on every spoken beat: I / actually / can (f5–28), a gap over the cut, read / text / the / distance. (f35–75), So / far (f75–95), only / the / left / eye / working. (f95–130), then nothing to the end. Words appear and vanish as hard swaps with no motion; the caption position and size never change even though the shots behind it change brightness and framing. The people in the footage are replaced here by blurred neutral plates.

Word-per-beat karaoke captioning: one word visible at a time as hard swaps at a fixed anchor, independent of and persisting across the picture cuts.

Phase notes

1. Frame 0 — Shot 1, no caption
2. Frame 5 — Words begin, one per beat
3. Frame 28 — Cut to shot 2; caption pauses
4. Frame 35 — Words resume across the cut
5. Frame 58 — Cut to shot 3
6. Frame 70 — Cut to shot 4
7. Frame 93 — Cut to shot 5
8. Frame 130 — Caption ends; picture holds

Use this example

1. Watch the source and draft at normal speed. Compare the order of actions, the held states, and the transition into the result.
2. Download the project and read its README before changing inputs. The package includes its implementation and declared dependencies.
3. Replace the sample content through the package's documented input or content builder. Keep source footage in the research comparison.
4. Render your adaptation and inspect text fit, framing, intermediate states, and entry and exit frames.

Recorded limits

• Visual fidelity is an authored draft judgement from sampled source frames; no independent blind review or native Resolve execution.
• Both variants rerendered from the packaged text renderer to repair a prior code/preview mismatch.

Inspect the original verification record ↗