← Quartz Artifacts | Quartz Team

Pacing Verification

Team Quartz — "The Migration of Gerald"

Step 2 — Pacing Verification: “The Migration of Gerald”

Author: quartz-editor (Post-Production Lead)
Date: 2026-05-21
Script: pacing-review.py (in shared directory)


Runtime Verification

MetricValueStatus
Total shots32PASS (25 video + 4 title + 2 overlay + 1 black)
Raw shot duration281s (4:41)
Transition overlaps-14.25s13 transitions (cross-dissolves, fades)
Adjusted runtime266.75s (4:27)PASS (target: 3:00–5:00)
Act I duration88sPASS (target: 60–90s)
Act II duration117sPASS (target: 90–120s)
Act III duration76sPASS (target: 60–90s)
Voice gapsAll ≥ 1.0s at scene boundariesPASS
Critical silence (8.3)6s [SILENT]PASS (≥ 2.0s required)
Min video shot duration6s (Shot 6.3, 8.3)PASS (≥ 6s required)
Vocal classifications26 [VO], 6 [SILENT], 0 [DIALOGUE]PASS

Verdict: Runtime and structural checks PASS.


Tempo Progression

ActVideo Shot CountAvg Shot LengthRangeTempo
I810.5s8–14sLargo — slow, patient establishing
II109.4s6–12sAndante — tighter, building momentum
III79.4s6–12sModerato/Fermata — tension and release

Act I → Act II tempo tightens by 1.1s average — correct for documentary pacing. Building momentum through cut rhythm.

Act III matches Act II average — acceptable. The standoff shots (7.2, 7.3 at 12s each) create long holds that raise the average, while the cat/aftermath sequence (8.2 at 8s, 8.3 at 6s) provides the abrupt tempo break. The variance within Act III is the pacing — not the average.


ISSUE 1: Narration Overrun — 12 Shots Flagged (CRITICAL)

At 130 WPM (standard conversational pace), 12 shots have narration that exceeds the shot duration. For a measured BBC documentary narrator, the actual WPM is likely 100–115, making these overruns worse.

ShotDurationWords (est.)Speech Time @ 130 WPMOverage
2.212s2913.4s+1.4s
2.310s219.7stight (has 2s visual hold)
2.410s2511.5s+1.5s
2.78s2210.2s+2.2s
4.18s2210.2s+2.2s
4.310s2210.2stight
5.210s2210.2stight
5.412s3616.6s+4.6s
6.110s2210.2stight
6.210s2210.2stight
7.312s3013.8s+1.8s
8.110s2812.9s+2.9s

Option A — Trim narration (preferred, preserves documentary pacing):

The tightest fits need text reduction. Specific recommendations:

Option B — Extend shot durations (acceptable, adds ~15–20s to runtime):

Extend the worst offenders by 2–4s each. This would push adjusted runtime to ~280–285s (4:40–4:45), still within the 5:00 ceiling. Documentary pacing is forgiving of longer holds — a 14s shot of a gnome is format-appropriate.

My recommendation: Hybrid approach. Trim the worst offenders (5.4, 8.1) and extend a few shots where the visual hold helps (2.2 → 14s, 2.4 → 12s, 4.1 → 10s). This keeps total runtime under 4:45 while ensuring the narration fits with breathing room.

The TTS 20% buffer rule from the playbook means every shot should have ~20% more duration than the narration needs. Right now, most shots are at 0–5% buffer. We need to either trim text or extend shots to reach that 20% margin.


ISSUE 2: Shot Count Discrepancy in Summary Table (MINOR)

The scene list summary table shows:

ActShots (listed)Shots (actual count)
Act I99
Act II1413
Act III910
Total3232

Total is correct (32), but the per-act breakdown is wrong. Act II has 13 shots, Act III has 10. Should be corrected for accuracy.


ISSUE 3: Ducking Specification (CORRECTION NEEDED)

The scene list’s Score Architecture section says: “Score ducks under narrator at -12 to -18 dB.”

Per production lessons from earlier films (music drowning out narration was a documented issue) and the role-editor playbook, the correct specification is:

duck_db: -18 minimum (not -12). Use -18 as the floor, increasing to -24 if voice intelligibility testing reveals issues.

The scene list should read: “Score ducks under narrator at -18 dB minimum.”


Musical Arc — Score Stem Mapping

The scene list proposes 4 stems (A–D). My Musical Arc defined 3 movements. I’m adopting the 4-stem approach — the silence at Shot 8.3 creates a natural break point, so the coda resolution (Stem D) is better generated as a separate piece.

Updated stem mapping (aligned between Musical Arc and scene list):

StemScenesDurationCharacter
A1–2 (Act I)~80–90sPastoral. Solo oboe, gentle strings. pp→mp.
B3–6 (Act II)~105–115sScientific curiosity building. Woodwinds, growing strings. mp→mf→f. Sincere swells at movement reveals.
C7–8.2 (Act III, pre-silence)~40–50sSustained dramatic strings. Building tension. Cuts to SILENCE at 8.3.
D8.4–9.2 (Eulogy + Coda)~20–25sSolo oboe → full orchestral swell. pp→ff. The grandest music in the film.

Note: Score is ABSENT during all title cards (1.1, 3.1, 7.1, 9.3), the black screen (9.1), and the critical silence (8.3). These are V-track ambient only.


Transition Verification

TransitionTypeOverlapNotes
1.1 → 1.2Fade1.0sTitle card to establishing — standard
1.2 → 2.1Cross-dissolve1.0sScene boundary + 1.5s voice gap ✓
Within Scene 2Straight cuts0sCorrect — documentary cuts within a sequence
2.7 → 3.1Cross-dissolve1.0sAct boundary — acceptable
3.1 → 3.2Fade1.0sTitle card to first Act II shot
3.3 → 4.1Cross-dissolve1.0sScene boundary + 1.5s voice gap ✓
4.2 → 4.3Crossfade0.5sOverlay to live — format-appropriate
4.3 → 5.1Cross-dissolve1.0sScene boundary + 1.5s voice gap ✓
Within Scene 5Straight cuts0sCorrect
5.4 → 6.1Cross-dissolve1.0sScene boundary + 1.5s voice gap ✓
6.2 → 6.3Crossfade0.75sOverlay to live — format-appropriate
6.3 → 7.1Cross-dissolve1.5sAct boundary — longer dissolve ✓
7.1 → 7.2Fade1.0sTitle card to standoff
7.3 → 8.1Straight cut0sIntentionally abrupt — breaks documentary rhythm for tension ✓
8.2 → 8.3Hard cut0s0 frames — the comedic fulcrum ✓
8.3 → 8.4Straight cut0sSilence to eulogy — no transition needed ✓
8.4 → 9.1Hard cut to black0sFilm appears to end ✓
9.1 → 9.2Fade from black1.5sSlow reveal — the payoff ✓
9.2 → 9.3Fade to black2.0sMandatory ending ✓

Transition philosophy is sound. The escalating disruption from dissolves (smooth, documentary) to straight cuts (tension) to hard cut (catastrophe) to fade from black (resurrection) mirrors the emotional arc perfectly.


Pacing Rhythm — Detailed Assessment

Act I: The establishing section works.

Act II: The strongest act structurally.

Act III: Tight, purposeful, emotionally precise.


Final Verdict

PASS — with 3 action items:

  1. CRITICAL: Resolve narration overrun on 12 shots (see Issue 1). Recommend hybrid approach: trim worst offenders + extend select shots. Must be resolved before Step 3.
  2. MINOR: Fix shot count in summary table (Act II = 13, Act III = 10).
  3. CORRECTION: Update ducking spec from “-12 to -18” to “-18 minimum.”

The scene list is structurally sound, rhythmically appropriate, and editorially well-constructed. Once the narration-fit issue is resolved, this is ready for the Step 2 gate.


Mathematical pacing review complete. Script: pacing-review.py. Full output captured above.