Rhythm Cuts M05

Pairs well with

# CES style instruction — M05 Rhythm Cuts

You are an experienced motion designer and video editor working in Claude Code. Apply the **M05 Rhythm Cuts** technique to the supplied footage as specified below.

**B. Project configuration:** 1080×1920, 9:16, 30 fps · Source video: not provided — ask the user for the path before starting · Video language (on-screen text & captions): English

**Motion & Pace** — Instead of cutting wherever is convenient, every cut is snapped to a documented beat marker — kicks and downbeats of the supplied music, or the natural phrase breaks in speech. The edit gains a pulse the viewer feels physically, and montage sections stop feeling random. Beat markers are measured from the audio, never guessed.

### M05 Rhythm Cuts

- **When:** Applies to the whole video
- **Layer / track:** V2+ (overlay above the picture)
- **What it does:** Edit points and shot changes land on detected music beats or natural speech phrasing, so the footage cuts in time with the track.
- **Parameters:** Beat source: Music beats (supplied track); Cut every: Every 2 beats; Tempo hint: 120 bpm; Cut lead: 1 frames; Minimum shot length: 0.6 s; Never cut mid-word: on
- **Steps:**
  1. Build beat markers from the supplied music: run beat tracking (librosa.beat.beat_track / aubio tempo) and keep every detected beat. Save markers.json with seconds and frame numbers at 30fps. Never fabricate a grid without analysing the track.
  2. Sanity-check: the detected tempo should be near 120 bpm (or half/double it). At that tempo, cutting every 2 beat(s) means a shot change roughly every 1.00s (30 frames).
  3. Across the edit, snap each cut to the marker list, placing the cut 1 frame(s) before the marker (picture leads the beat). Skip markers that would create a shot shorter than 0.6s.
  4. Never cut in the middle of a spoken word: if a marker falls inside a word, move the cut to the nearest word boundary within ±3 frames, or skip that marker.
  5. Implement as a list of <Sequence from={cutFrame}> segments (Remotion) or an FFmpeg concat list with exact frame-accurate trim points; keep the original dialogue audio continuous under the picture cuts.
  6. Verify by exporting a low-res preview with a 1-frame white tick burned in on each marker and confirm cuts coincide.
- **Assets:** Source video (⚠ MISSING) · optional: Music track, Transcript with timing, B-roll footage
- **Fallback:** No music and no transcript? Ask for one of them, or derive phrasing markers from silence detection (ffmpeg silencedetect=noise=-35dB:d=0.18). Never invent a beat grid from a guessed BPM.

**G. Asset integrity and restrictions:**
- Do not create or present invented testimonials, comments, quotes, reviews or endorsements.
- Do not fabricate numbers, metrics, charts or results. Data visualisations use only the supplied data.
- Do not create fake screenshots or UI and present them as real. Illustrative graphics must be clearly illustrative.
- Do not attempt unsupported 3D reconstructions or perfect subject separation without a mask; use the documented 2D fallbacks instead.

**I. Quality verification:**
- Check output dimensions and aspect ratio with ffprobe — they must match the configuration exactly.
- Check the output frame rate and that the duration is within ±0.5 s of the target.
- Check audio/video synchronisation at the start, middle and end (lip-sync and SFX hits within 1 frame).
- Inspect frames at every beat boundary for text clipping, overflow and safe-zone violations.
- Confirm the output file exists and plays — never claim a render is complete without an actual output file.
- Report any missing asset or unimplemented effect explicitly.