01Tier 1 forks — autonomous decisions (no human user)
| # | Fork | Decision | Rationale |
|---|---|---|---|
| 1 | Production Target | Higgsfield Soul Cinema (soul_cinema_studio) for speaker; Higgsfield Marketing Studio for the lone mechanism B-roll; post-production overlay for ALL on-screen text (Soul renders text unreliably per avatar_tool_prompt_syntax.md) | Solo × at-home × PROVEN speaker = Soul Cinema is the tool-choice-matrix recommendation. Speaker 1.1 is the Sparks portfolio's most-tested avatar lineage — Soul produces highest-fidelity face for the kitchen-table-warmth register. Heygen would over-polish; Marketing Studio is for product beats. |
| 2 | Visual Style Direction | Lived-in kitchen-table intimacy (Anchor A + B blend): warm daylight from a single window-left source, walnut surface + plant + minimal background dressing, sustained wide-medium framing for the body, single cut to close-up only on the "your body isn't broken" beat per the script's Field 21 editor note | Two of the top 5 anchors are this exact look (Resilia Amish-grandmother + Happy Mammoth kitchen). Solo × at-home's V1-canonical aesthetic is the empirical winner; the brand bible §11 "kitchen-table warmth" + speaker file's "lectern, simple direct-to-camera with a plant or bookshelf behind her" both lock it. No reason to invent a new direction when the canonical lane is what the speaker is PROVEN in. |
| 3 | B-roll Source Mix | Mostly NO B-roll (talking-head heavy per Anchor A + script's "minimal cuts" Field 21); ONE AI-generated mechanism cutaway at Beat 3 (sleeping woman → cortisol/glucose visual overlay); NO product B-roll at Beat 5 (we name the brand verbally, on-screen text carries the URL) | Script's Field 21 says "Podcast-cut. Minimal cuts. Wide-medium seated framing. Cut to close-up only on..." Honor it. Anchor A is 4 minutes of nearly pure talking-head — proves the format works. Adding B-roll density would fight the speaker's authority. The single mechanism cutaway is earned (mechanism is the script's highest-stakes beat per universal Rule 16). |
| 4 | Mute-test Asset Strategy | Strategy 1 — Speaker face + on-screen text (face direct-to-camera; on-screen text reveals "3 in the morning. Heart slamming. Sense of doom." kinetic stagger on the hook's three specificity beats) | Kitchen-table-warmth speaker's face IS the trust anchor — 70yo natural-health authority is the audience's most-trusted archetype on sight per speaker file ("The audience already trusts this archetype on sight"). External imagery would compete. Anchor A (cornerstone) opens with face + text-OCR captions — same playbook. |
| 5 | Comp Library Inspiration Depth | Pattern-inspired (technique not asset) | Solo × at-home is V1's most-tested combo — every move has comp depth. Pattern-borrow preserves the script's "3:17am" specificity as the differentiator. Full-pattern-borrow would echo Anchor A's Amish frame; combinatorial blending of 5 anchors would muddy a script designed for one continuous voice. |
02Comp library inspiration anchors
[Anchor A] 1142646044918657 — resilia — "What nobody tells you about living past 90" — winner_score 51.89 view ad
VIDEO; CORNERSTONE — only direct solo × at-home × 70+-woman-authority match in pool. 4-minute yapper. Older woman in rustic indoor setting, slow-burn folk-wisdom narrative. Single-speaker, talking-head only, sustained wide-medium framing throughout. Slow-paced text-on-screen captions reinforce stress beats. Tagged talking-head-only + on-screen-captions-heavy + slow-burn in creative_techniques. ANCIENT — Generational Heritage framework.
- Hook pattern: "What nobody tells you about [outcome]" — soft authority-confession opener at slow cadence; speaker face + on-screen text-OCR captions in first 5s. Mute test = face + text caption stacked.
- On-screen text style: Persistent stress-beat captions reinforcing what the speaker says verbally; clean sans-serif lower-third + center-frame stagger on enumeration; ≤8 words per appearance; held 3-4s.
- B-roll cuing: Effectively NONE for the body of the script — talking-head sustains. This is the format-proof point: a single warm voice carries 4 minutes without B-roll density.
- Transition style: Soft cuts (jump cuts within a single sustained frame, podcast-aesthetic). No hard pivots between sections.
- Mute-test asset: Speaker face + text-OCR caption — both survive past t=5s.
- Extractable for our script: the WHOLE template — talking-head minimal-cut + on-screen text reinforcing stress beats + slow-burn cadence. Single-speaker register sustained without B-roll padding. This is what "solo × at-home + podcast-cut + minimal edits" looks like in a winning ad.
[Anchor B] 1874728453166829 — happymammoth — "We need to talk about menopause… honestly." — winner_score 57.20 view ad
VIDEO. Kitchen-confessional UGC. Tagged kitchen-confessional + talking-head-only + pattern-interrupt-vocal-shifts. SHAPE — Peer-to-Peer Symptom variant. Different register from our speaker (UGC peer, not 70yo authority) but identical setting.
- Hook pattern: "We need to talk about [topic]… honestly." — permission-frame soft opener. Direct-to-camera, intimate kitchen framing.
- On-screen text style: Lower-third bullets on benefit-stack moments; speaker carries hook, text comes later for proof.
- B-roll cuing: Light — single product unboxing/demo beat integrated mid-script. Otherwise talking-head.
- Transition style: Vocal shifts ARE the pattern interrupt — pacing/tone changes mid-take handle what a cut would otherwise carry. No editorial cuts during the speaker's monologue.
- Mute-test asset: Speaker face + kitchen background + (post-production) on-screen text. The kitchen IS visible mute-test context — "this is a real woman talking honestly in her real kitchen."
- Extractable for our script: the setting credibility — visible kitchen detail (cookbook, plant, real cabinet light) signals "real home, not set." Steal the principle, not the shot: our 70yo authority gets her kitchen visible the same way.
[Anchor C] 759146623860120 — resilia — "How often to take oil of oregano? 🌿 From a real cancer researcher" — winner_score 72.07 view ad
VIDEO. HIGHEST video winner_score in pool. Talking-head authority + dosing-reframe template. scripted-authority voice register. Different niche but TEMPLATE-transferable for our mechanism reveal beat.
- Hook pattern: Authority-credential question-frame. "How often [X]? From a real [credential]." Credential drop in first 3s.
- On-screen text style: Pattern-interrupt text overlay reinforces credential; ingredient specificity overlays on mechanism beat.
- B-roll cuing: Sparing — uses ingredient/product close-ups on mechanism deep-dive; otherwise talking-head.
- Transition style: Educational-reframe pacing — vocal acceleration on the "actually, the cycling myth is wrong" reveal; visual stays locked.
- Extractable for our script: the mechanism-explanation cadence + ingredient-specificity overlay on Beat 5 (Iodine + Selenium + Zinc + Copper enumeration). Authority delivers the named-ingredient line; on-screen text stagger reinforces each name as it's spoken.
[Anchor D] 1465022601769793 — trysoluma — "If you've been waking up at exactly 3am every night..." — winner_score 56.60 view ad
STATIC IMAGE; thematic-bullseye but format-mismatch. Hook STRUCTURE transferable only — NOT visuals. PASTOR — Star-Story-Solution variant. The exact 3am cortisol audience our script targets; VOC validation that this audience exists and clicks (specificity tier 5).
- Extractable for our script: Hook structure validation — "If you've been waking up at exactly [time] every night... your doctor keeps offering [drug A or B]... and [tried-list] does absolutely nothing" — IS the audience-call pattern Laurie responds to. Our hook already mirrors this structure (Field 15: "If you bolt awake at three in the morning..."). NOT borrowing the static-image format or anatomical illustration; using as confirmation our hook lives in proven territory.
[Anchor E] 887515617080742 — tryjevi — "I'm cortisol you need me in the morning" — winner_score 45.94 view ad
VIDEO; 3D Pixar-animation musical. Format-mismatch but mechanism-overlay TECHNIQUE transferable. DOMINO — Personified Villain variant.
- Extractable for our script: Just one element — the personified cortisol/adrenaline mechanism cutaway as a possible reference for our Beat 3 mechanism B-roll. But we will NOT use musical/animation register (would shatter the 70yo authority frame). What we steal: the mechanism-as-character story-beat — "your liver runs out of glucose, your body panics, dumps adrenaline" is a personified-villain microbeat. Visual stays on speaker; voice carries the personification.
03Per-beat storyboard
▸ Beat 1: HOOK — strange-symptom specificity + permission frame + open Loop 1 0:00–0:18
| Persuasion goal | Mute-test stop-scroll asset; specificity-anchored credibility-open (3am specificity + the three-pain stack); open Loop 1 ("what's actually happening at 3am"); permission frame ("I want you to listen to this"); introduce Laurie as the anchor character. |
| Shot type | Wide-medium seated framing, eye-level. Speaker visible from waist up at a kitchen table; warm-lit kitchen counter + window-left light + plant in soft-focus behind. Slight handheld feel (locked but with subtle organic micro-drift, NOT shaky). |
| Camera | Eye-level, locked, centered. Speaker positioned slightly camera-right so left negative space holds the window-light source visible. Shallow depth-of-field — plant + bookshelf soft-focus behind. NO zoom or push. Wide-medium sustained — this is podcast-cut, not cinematic cut. |
| On-screen text | Kinetic stagger reinforcing the three specificity beats from the hook line. "3 in the morning." — lower-third, sans-serif, cream-on-translucent-dark, letter reveal locked by t=1.5s, hold to t=4.5s. "Heart slamming." — fade-cross at t=4.7s to t=8.5s. "Sense of doom." — fade-cross at t=8.7s to t=12.5s. Then clear t=12.5s onward (lets the aside breathe). All ≤4 words each (VS2 ✓). Sans-serif, NOT italic — matches kitchen-table register (italic would over-stylize). |
| B-roll cue | NONE. Speaker face + on-screen text carries the entire 18s. Per Anchor A's sustained-talking-head model. B-roll here would undercut the script's "Sit with me a minute" intimacy. |
| Lighting | Default at-home setting per setting file — warm daylight from kitchen window camera-left at ~5500K (natural-window-light look — speaker file: "natural-window-light look is canonical. Too-perfect lighting reads as set"). Slight imperfection (one slightly blown highlight on the window frame, slight shadow camera-right) reads as authentic. Edison-bulb practical OFF — we want daylight only. |
| Transition out | Soft cut at t=0:18 to medium (still seated, same angle, no jump). Podcast-aesthetic micro-cut, NOT a hard pivot. |
Avatar-tool prompt (Higgsfield Soul Cinema)
MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "70-year-old woman with silver chin-bob hair, soft warm expression, slight forward lean toward camera, seated at a worn walnut kitchen table, wearing a sage-green button-up cardigan over a cream blouse, hands resting clasped softly on the table, slight conspiratorial intimacy in the eyes, warm natural daylight from a kitchen window camera-left, plant in soft-focus behind her shoulder, cookbook stack and small ceramic mug on counter background blurred, walnut + cream + sage palette, wide-medium seated framing, eye-level locked camera, shallow depth-of-field, cinematic-warm 5500K natural-window-light, authentic kitchen detail visible but unfussy, no studio polish, slight imperfection in lighting reads as real home" NEGATIVE_PROMPT: "text, multiple faces, distorted hands, cluttered background, studio backdrop, fluorescent lighting, over-polished, beautified skin, jewelry, prints on clothing, podcast-studio cues, modern minimalist white kitchen" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level wide-medium, no movement, shallow depth-of-field DURATION (image-to-video): 18s (single sustained take — break into 2 sub-renders if Soul drift kicks in past 8s; composite seamless in post) ON_SCREEN_TEXT: handled in post-production overlay — see "On-screen text" field above for exact stagger spec B_ROLL: none
| Mute-test | PASS — face + on-screen text-stagger carries full 18s sound-off. VH1 ✓ / VS2 ✓ (≤4 words per appearance) / VS5 ✓ (asset survives full 5s; survives full 18s). |
| Production-limitation cross-ref | Soul drift on a single 18s render is the known Soul limitation. Workaround: break into 2 × 9s renders, composite seamlessly via post (a real natural blink-cycle at the cut point hides it). No HARD flag — within current Soul capability. Note: per production_limitations.md heuristic, doc is <3 non-template entries (Phase 1 state), so YS6 cross-check silently skips. |
| Comp library inspiration | [Anchor A — CORNERSTONE] sustained talking-head with on-screen text reinforcing stress beats. [Anchor D — STRUCTURE ONLY] hook STRUCTURE validation ("waking up at exactly 3am" audience-call). Specifically NOT borrowing Anchor D's static-image format or anatomical illustration aesthetic. [Anchor B] kitchen-setting credibility (real cookbook, real plant, real window-light). |
▸ Beat 2: VALIDATION + doctor/husband-dismissal pivot + open Loop 2 0:18–0:38
| Persuasion goal | Validate Laurie's failed-solutions stack (melatonin / magnesium / weighted blanket / trazodone) — this is the universal craft Rule 4 "and then it got worse" + Failed Solutions Stack (Y6 surgical fragment); the doctor-said/husband-said pivot closes the validation loop; aside #1 ("wired but tired") delivers Y2 warmth glue at 30s mark per yapper cadence floor; opens Loop 2 ("what nothing in your medicine cabinet touched the real cause of"). |
| Shot type | Same wide-medium seated framing — visual continuity per VH2. No cut to close-up here (per script Field 21: cut to close-up only on the "your body is screaming for help" beat in Beat 3). |
| Camera | Eye-level, locked, centered. Same composition as Beat 1. Speaker may visually shift weight slightly into the failed-solutions list (organic body language, not a directed move). |
| On-screen text | Failed Solutions Stack reinforced visually — "Melatonin." "Magnesium." "Weighted blanket." "Trazodone." — center-frame stacked stagger, ~0.7s reveal interval, starting at t=0:19, completed by t=0:22, then cross-fade out at t=0:24 (lets the "None of it touched the 3am wake-up" land sound-on). Four words total per beat — VS2 ✓. Then "Wired but tired." lower-third — fades in at t=0:32 (aligned with the aside), hold to t=0:36. Italics OK on this aside line — it's a quoted audience-VOC phrase. |
| B-roll cue | NONE. Talking-head sustains per Anchor A discipline. The Failed Solutions Stack would be tempting to B-roll (pill bottles, blanket, prescription pad) — but B-roll-stacking the list would undercut its surgical fragment-density punch per Y6. The on-screen text stagger IS the visual cadence; B-roll would compete. |
| Lighting | Sustained at-home default per VH2. |
| Transition out | Soft cut at t=0:38 — micro-cut into the mechanism reveal beat. NOT the script's earned hard pivot (that lives later — between Beat 4 and Beat 5). |
Avatar-tool prompt (Higgsfield Soul Cinema)
MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old woman, same walnut kitchen table seated, same sage-green cardigan + cream blouse, same warm window-left daylight, expression shifts to slightly knowing-tired (validation expression — she's seen this pattern hundreds of times), slight head-tilt on the 'wired but tired' aside, hands stay clasped or one moves slightly to gesture during the failed-solutions list, walnut + cream + sage palette continuity, wide-medium seated framing identical to prior beat, eye-level locked camera, shallow depth-of-field, natural-window-light 5500K, kitchen detail unchanged" NEGATIVE_PROMPT: "text, location shift, lighting shift, wardrobe shift, dramatic gesture, animated/exaggerated expression, podcast-studio cues" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level wide-medium, no movement DURATION (image-to-video): 20s (likely 2 × 10s sub-renders for Soul stability) ON_SCREEN_TEXT: handled in post per stagger spec above B_ROLL: none
| Mute-test | N/A (non-hook beat). However on-screen text-stagger of failed-solutions list passes mute test independently — a sound-off scroller who lands here still gets the value. |
| Production-limitation cross-ref | Same Soul drift workaround applies (split renders). No flags. |
| Comp library inspiration | [Anchor A] sustained-talking-head with caption reinforcement. [Anchor B] kitchen-table register continuity. NOT borrowing Anchor E's personification (would shatter the validation register). |
▸ Beat 3: MECHANISM REVEAL — closes Loop 1 — the "your body isn't broken" close-up beat 0:38–1:05
| Persuasion goal | Mechanism reveal — T4→T3 conversion gap → low T3 → liver stops storing glucose overnight → 3am adrenaline dump. Closes Loop 1. The script's HIGHEST-stakes beat (per universal master prompt Rule 16 — mechanism paragraphs typically the highest-stakes paragraphs in the script). The "Your body isn't broken. Your body's doing its job — keeping you alive at 3am" aside is the script's signature reframe beat — Field 21 explicitly directs the cut to close-up here. Validation-reframe ("the melatonin couldn't touch it because melatonin isn't the problem. The problem is upstream") delivers the persuasion-arc pivot. |
| Shot type | Wide-medium for t=0:38–0:52 (T4→T3 explanation); CUT to close-up at t=0:52 for the "Your body isn't broken" aside — held through t=0:57; then back to wide-medium at t=0:57 for the "melatonin couldn't touch it" close. ONE optional cutaway: sleeping-woman + cortisol/glucose-curve overlay mechanism cutaway at t=0:48–0:52 (4 seconds) — strategically the highest-leverage cutaway in the script. |
| Camera | Beat 3a (0:38–0:48): wide-medium locked, same as Beat 2. Beat 3b (0:48–0:52): cutaway B-roll (no camera on speaker). Beat 3c (0:52–0:57): CLOSE-UP — tighter framing, head + shoulders, eye-level, locked, NO push-in (a slow push-in here would over-direct; the cut itself is the cinematic move). Beat 3d (0:57–1:05): return to wide-medium for the close. |
| On-screen text | Mechanism enumeration reinforces the spoken biology — "T4 → T3" lower-third clinical-translation overlay fade-in at t=0:42, hold to t=0:46. "Selenium + Zinc to flip the switch." center-frame at t=0:47, hold to t=0:51. During the cutaway (t=0:48–0:52): NO additional text (B-roll carries). Then on close-up (t=0:52–0:57): NO text — the close-up + speaker face is the entire mute-test asset; competing text would dilute the reframe beat. Then at t=1:00 "The problem is upstream." center italic lowercase, hold to t=1:05. All ≤7 words — VS2 ✓. |
| B-roll cue | t=0:48–0:52 — 4-second mechanism cutaway. Soft, painterly, NOT clinical-3D-medical. Asset description: Quiet visual of a woman sleeping, then a translucent overlay graphic shows a smooth glucose curve descending through the night, with a small adrenaline-spike at the curve's bottom — restrained, gentle illustration aesthetic (think watercolor + line, NOT Anchor E's 3D Pixar animation, NOT Anchor A's vintage-anatomical aesthetic — gentler than both because the speaker's voice is the authority, the visual is supporting). Single 4s sustained shot, slight Ken Burns push. VS3b per-cue citation: [Anchor E — TECHNIQUE ONLY] — borrowing the mechanism-as-overlay TECHNIQUE (visualization on mechanism reveal beat), explicitly NOT borrowing Anchor E's musical-animation register (which would shatter the 70yo authority frame). |
| Lighting | Beat 3a/3c/3d at-home default sustained. Beat 3b (cutaway): painterly soft-light overlay, dim blue-grey palette to evoke "3am bedroom" contrasted against the warm kitchen — palette contrast is intentional, returns to warm at t=0:52 with the close-up cut. Beat 3c close-up: same key light from window-left, slightly tighter framing makes the light fall feel more intimate. |
| Transition out | Beat 3a → 3b: cross-dissolve (soft, 0.3s) into mechanism overlay — the dissolve telegraphs "we're entering the biology." Beat 3b → 3c: hard cut from cutaway back to close-up — this hard cut IS the visual translation of "Your body isn't broken." Beat 3c → 3d: soft cut back to wide-medium. Beat 3d → Beat 4 (next): soft cut. |
Avatar-tool prompt (Higgsfield Soul Cinema, multi-scene)
=== SCENE 3a (t=0:38–0:48 — mechanism explanation, wide-medium) === MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old natural-health authority, same walnut kitchen table seated, sage-green cardigan + cream blouse, warm window-left daylight, EXPLAINING expression — slight forward lean, gestures one hand softly on first 'T4' then 'T3' to telegraph the conversion-flip, knowing-warm eyes, voice carries the biology, walnut + cream + sage palette, wide-medium seated framing eye-level locked, shallow depth-of-field, 5500K natural-window-light, kitchen continuity" NEGATIVE_PROMPT: "text, dramatic gesture, lecture-podium pose, clinical register, location shift, lighting shift" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level wide-medium, no movement DURATION (image-to-video): 10s ON_SCREEN_TEXT: handled in post per spec above === SCENE 3b (t=0:48–0:52 — mechanism cutaway B-roll) === TOOL: Higgsfield Marketing Studio (image-to-video) — separate generation PRESET: lifestyle_scene SCENE_PROMPT: "Painterly soft-light illustration: a woman in her 50s sleeping in bed at 3am, partial silhouette under blanket, dim blue-grey bedroom palette, then a translucent watercolor overlay shows a smooth glucose curve descending across the night with a small upward adrenaline-spike at the curve's bottom right, gentle line + watercolor aesthetic, NOT 3D Pixar animation, NOT vintage clinical anatomical illustration, gentle painterly mechanism visualization, no medical jargon labels on screen, just the curve and the figure, slow Ken Burns push-in across 4 seconds, no human voice or sound, soft palette: dim navy/grey for bedroom + warm amber accent for the adrenaline spike" NEGATIVE_PROMPT: "3D animation, Pixar style, cartoon, medical diagram with labels, anatomical cross-section, vintage scientific illustration, text overlays, clinical register, animated character" ASPECT_RATIO: 9:16 DURATION: 4s ON_SCREEN_TEXT: none during this cutaway === SCENE 3c (t=0:52–0:57 — CLOSE-UP — "Your body isn't broken" aside) === MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old natural-health authority, CLOSE-UP framing — head and shoulders fill the frame, eye-level locked camera, soft expression — direct eye-contact with viewer, slight reassuring smile that doesn't read as performed, sage-green cardigan visible at shoulders, warm window-left key light slightly more intimate at this framing, walnut + cream + sage palette continuity, kitchen background blurred to soft bokeh behind shoulders, 5500K natural-window-light, NO movement — single sustained close-up" NEGATIVE_PROMPT: "text, push-in, zoom, dramatic gesture, lecture pose, performed smile, location shift, lighting shift, hands in frame (close-up is head+shoulders only)" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level CLOSE-UP, no movement (DO NOT push in — the cut itself is the cinematic move) DURATION (image-to-video): 5s ON_SCREEN_TEXT: none — speaker face + reassuring aside is the entire mute-test asset on this beat; competing text would dilute === SCENE 3d (t=0:57–1:05 — return to wide-medium for close) === MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old natural-health authority, back to wide-medium seated framing, walnut kitchen table, sage-green cardigan + cream blouse, warm window-left daylight, resolved-confident expression, slight head-shake on 'melatonin isn't the problem' (subtle), authority cadence on 'upstream', kitchen continuity, eye-level locked, shallow depth-of-field, 5500K" NEGATIVE_PROMPT: "text, push-in, zoom, location shift, lighting shift" ASPECT_RATIO: 9:16 DURATION (image-to-video): 8s ON_SCREEN_TEXT: "The problem is upstream." — center italic lowercase — fade in t=1:00, hold to t=1:05 (handled in post)
| Mute-test | N/A (non-hook). However the cutaway + caption at t=0:48–0:52 passes mute test as a SECONDARY scroll-recapture asset for any viewer landing mid-script. Close-up at t=0:52–0:57 passes mute test independently — speaker face direct-eye-contact + the warm reframe carries even without sound (the visual itself reads "she's reassuring me"). |
| Production-limitation cross-ref | Soul rendering reliable for sustained head-and-shoulders close-up. Mechanism B-roll generated separately via Marketing Studio — well within tooling (painterly lifestyle scene). One info-note: the 0.3s cross-dissolve from speaker to cutaway requires a clean cut-point in the speaker render — Soul renders should leave a natural blink/pause at the dissolve frame. No flags. |
| Comp library inspiration | [Anchor A — primary] sustained talking-head as default, with on-screen text reinforcing the biology terms. [Anchor C] ingredient-specificity overlay technique on "Selenium + Zinc" enumeration. [Anchor E — TECHNIQUE ONLY per VS3b] mechanism-overlay cutaway PATTERN borrowed (not the 3D Pixar register — adapted to gentle painterly watercolor that doesn't fight the speaker's authority). The close-up cut at the "your body isn't broken" beat is original-direction — rationale: Field 21 explicitly directs it and it's the script's signature reframe beat; no comp anchor has this exact tonal move (Anchor A's slow-burn never cuts to close-up; Anchor B's UGC peer doesn't have an authority-reassurance equivalent). VS3b ✓ per-cue. |
▸ Beat 4: Closes Loop 2 + cortisol third leg 1:05–1:25
| Persuasion goal | Close Loop 2 — the cortisol leg (Reverse T3 fake-keys-jamming-the-locks). Adds the third mechanism beat (selenium + zinc + ashwagandha story setup) that makes the unfair advantage land in Beat 5. The "without something to settle the cortisol — the stress wins" line is the bridge per Y7 to the solution introduction at Beat 5. |
| Shot type | Wide-medium seated framing — visual continuity per VH2. NO close-up cuts here (per Field 21: close-up only on Beat 3's reframe beat). |
| Camera | Eye-level, locked, centered. Same composition as Beats 1/2/3a/3d. |
| On-screen text | "Reverse T3" lower-third overlay fade-in at t=1:10, hold to t=1:14 — clinical-translation reinforcement. "Fake keys jamming the locks." center italic at t=1:16, hold to t=1:21 — the body-metaphor mechanism phrase per Rule 16 (and the speaker file's signature phrase: "fake keys jamming the locks" is documented as a 1.1 Natural-Health Authority body-metaphor). 5 words — VS2 ✓. |
| B-roll cue | NONE. Speaker carries the cortisol-reverse-T3 logic on voice alone. Adding visual here would over-stack mechanism information (the Beat 3 cutaway is the script's one earned mechanism visual). |
| Lighting | Sustained at-home default. |
| Transition out | Soft cut at t=1:25. The Y7 earned hard pivot lives between Beat 4 and Beat 5 — but it's expressed verbally ("That's why what I give the women in my practice is three things, not one") not as a hard visual cut. Soft cut preserves the kitchen-table register. The hard pivot is in the speaker's voice (resolved confidence shift), not the editor's cut. |
Avatar-tool prompt (Higgsfield Soul Cinema)
MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old natural-health authority, walnut kitchen table seated, sage-green cardigan + cream blouse, warm window-left daylight, EXPLAINING-with-conviction expression, slight emphasizing gesture on 'fake keys jamming the locks' (one hand briefly mimics a key-turning motion, very subtle), then return to clasped, walnut + cream + sage palette, wide-medium seated framing eye-level locked, shallow depth-of-field, 5500K natural-window-light, kitchen continuity" NEGATIVE_PROMPT: "text, dramatic gesture, location shift, lighting shift, podcast-studio backdrop" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level wide-medium, no movement DURATION (image-to-video): 20s (likely 2 × 10s sub-renders) ON_SCREEN_TEXT: handled in post per spec above B_ROLL: none
| Mute-test | N/A (non-hook). On-screen text "Fake keys jamming the locks." passes mute test as scroll-recapture mid-script — strong body-metaphor caption. |
| Production-limitation cross-ref | No flags. |
| Comp library inspiration | [Anchor A] sustained-talking-head + clinical-term lower-third pattern. [Anchor C] mechanism-explanation cadence (authority voice carries the biology, text reinforces the named term). |
▸ Beat 5: UNFAIR ADVANTAGE + DAMAGING ADMISSION + Kindled product reveal 1:25–1:45
| Persuasion goal | Unfair Advantage = the three-thing formula (Selenium + Zinc + Ashwagandha + supporting cast). Damaging admission ("Won't work overnight. Talk to your doctor — especially if you have Hashimoto's"). Kindled product reveal (verbal name-drop, on-screen text carries URL). This is the one earned hard pivot per Y7 — handled verbally not editorially. The "hypothyroid guts are slow guts" aside is the Y6 surgical fragment for this beat. |
| Shot type | Wide-medium sustained — visual continuity. NO product B-roll insert (per Tier 1 Fork 3 decision — script's Field 21 directs minimal cuts; product B-roll would shatter the kitchen-table register; the brand name is verbal + the URL on-screen handles the visual recognition). |
| Camera | Eye-level, locked, centered. Same wide-medium framing. |
| On-screen text | Ingredient stagger on the formula enumeration — "Selenium + Zinc." lower-third at t=1:28, fade to t=1:32. "Iodine + Copper." lower-third at t=1:33, fade to t=1:36. "Ashwagandha." lower-third at t=1:37, fade to t=1:40. Each ≤4 words — VS2 ✓. Then product name reveal — "KINDLED" brand text lower-third bold sans-serif (NOT all-caps — the brand reads cleaner as title-case "Kindled") at t=1:41, hold to t=1:45. Then "trykindled.com" smaller URL on the same line — fade-in below the brand name at t=1:43, hold to t=1:45 (persists into Beat 6). |
| B-roll cue | NONE. No Kindled bottle B-roll — the script's Field 21 directs minimal cuts; product B-roll would over-direct what is a deliberately understated reveal in a kitchen-table register. Verbal name-drop + on-screen URL handles brand recognition. VS3b per-cue rationale: "no B-roll" is itself a decision per anchor — Anchor A sustains 4 minutes without product B-roll and converts; we follow the cornerstone's discipline. The grey-zone playbook's adjunct framing lives in voice ("Talk to your doctor — especially if you have Hashimoto's"). |
| Lighting | Sustained at-home default. |
| Transition out | Soft cut at t=1:45 to Beat 6 (CTA close). Speaker holds same framing. |
Avatar-tool prompt (Higgsfield Soul Cinema)
MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old natural-health authority, walnut kitchen table seated, sage-green cardigan + cream blouse, warm window-left daylight, RESOLVED-CONFIDENT expression — she's arriving at her recommendation, slight conversational gesture during the three-ingredient enumeration (one finger raise per ingredient counting, very subtle), then a small head-shake on 'hypothyroid guts are slow guts' (authority-aside delivery), softens on the Hashimoto's caveat, walnut + cream + sage palette, wide-medium seated framing eye-level locked, shallow depth-of-field, 5500K natural-window-light, kitchen continuity" NEGATIVE_PROMPT: "text, dramatic gesture, product-holding pose (no product in hand), location shift, lighting shift, salesy expression, podcast-studio backdrop" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level wide-medium, no movement DURATION (image-to-video): 20s (likely 2 × 10s sub-renders) ON_SCREEN_TEXT: handled in post per spec above B_ROLL: none — verbal brand-reveal only
| Mute-test | N/A (non-hook). On-screen text — ingredient stagger + brand reveal + URL — passes mute test for sound-off scrollers landing here. |
| Production-limitation cross-ref | No flags. |
| Comp library inspiration | [Anchor C — primary] ingredient-specificity overlay technique (each named ingredient gets its own lower-third reveal as it's spoken). [Anchor A] sustained-talking-head discipline (no product B-roll at the reveal beat — speaker authority carries the recommendation). [Anchor B] kitchen-confessional integrated product reveal pattern (verbal name-drop in the natural flow of the conversation, no unboxing). DAMAGING ADMISSION beat is original-direction — rationale: speaker file's "Honest about limitations" voice principle + brand bible §11 ("Honest about limitations" is a brand voice principle); no comp anchor cleanly transfers because most comp ads don't hard-cap their own claim with a Hashimoto's caveat. VS3b ✓ per-cue. |
▸ Beat 6: CTA close + identity restoration 1:45–2:00
| Persuasion goal | Soft CTA close. Patricia's identity-restoration story ("slept through the night for the first time in five years a couple months in") delivers the after-state visualization (Rule 8 future pacing). "Said she felt like herself again" — the universal Mass Desire #1 from brand bible §6 ("I want my old self back"). URL repeat closes the script. |
| Shot type | Wide-medium sustained — visual continuity through to the close. NO close-up here either (already used the one at Beat 3 per Field 21 directive; over-using close-ups dilutes the Beat 3 cut). |
| Camera | Eye-level, locked, centered. At t=1:55, very slight micro-push-in (NOT a zoom — barely perceptible, just shifting the speaker fractionally larger in frame) to telegraph the CTA-direct-address shift from "Patricia's story" to "you, the viewer." This is a subtle move — Soul can struggle with push-ins, so the alternative is just a perfectly held wide-medium with NO movement. Default to no movement. |
| On-screen text | "trykindled.com" persistent lower-third URL — held from t=1:45 (continuing from Beat 5) through end at t=2:00 + 2s post-fade. At t=1:55 add "Said she felt like herself again." center italic on cream-translucent — hold to t=1:59, fade. 6 words — VS2 ✓. Final frame holds URL at center-screen for 2s post the speaker's final word (allows viewer time to absorb). |
| B-roll cue | NONE. Y8 — CTA stack at close always present in solo dynamic. Speaker face + URL overlay carries the close. No product B-roll, no Patricia stock-image (Patricia is a verbal character — a stock-image insert would shatter the credibility). |
| Lighting | Sustained at-home default. Slight key-light intensity bump (~5%) at t=1:55 — barely perceptible, telegraphs "warmth" on the close. |
| Transition out | N/A — script ends. Speaker holds final frame with subtle natural expression (slight smile, satisfied-warm) for 1s post-final-word. Then fade to soft cream-on-walnut card with persistent "trykindled.com" URL center-screen for 2s. |
Avatar-tool prompt (Higgsfield Soul Cinema)
MODEL: soul_cinema_studio SOUL_ID: kindled_speaker_1_1_natural_health_authority_v1 PROMPT: "Same 70-year-old natural-health authority, walnut kitchen table seated, sage-green cardigan + cream blouse, warm window-left daylight, WARM-SATISFIED expression on the Patricia story (the rare not-performed smile from a practitioner remembering a real patient outcome), softens further on 'said she felt like herself again', resolves to gentle direct-eye-contact for the URL close, walnut + cream + sage palette, wide-medium seated framing eye-level locked, shallow depth-of-field, 5500K natural-window-light, kitchen continuity, final frame holds gentle satisfied expression for 1s post-final-word" NEGATIVE_PROMPT: "text, push-in (default no movement), dramatic gesture, performed smile (must read as remembering real patient, not advertising), location shift, lighting shift, salesy expression, urgency cues" ASPECT_RATIO: 9:16 CAMERA_DIRECTION: locked eye-level wide-medium, no movement (default; optional very-subtle 1-2% push at t=1:55 if Soul handles it cleanly — otherwise skip) DURATION (image-to-video): 15s + 1s held final frame ON_SCREEN_TEXT: handled in post per spec above; URL persists 2s past final frame on a soft cream-on-walnut card B_ROLL: none POST-SCRIPT CARD (t=2:00–2:02): PROMPT (Higgsfield Marketing Studio — closeup_product preset OR brand asset library): "Cream off-white background with subtle walnut/wood texture overlay, centered text 'trykindled.com' in warm dark-charcoal sans-serif, no product image, no logos beyond the URL, minimalist warm-kitchen palette card, no graphics or borders, calm dignified closing card consistent with the kitchen-table register, 2 second hold" DURATION: 2s
| Mute-test | N/A (non-hook). However the persistent URL + closing italic caption + speaker's satisfied close passes mute test for sound-off viewers — the URL is the primary action signal. |
| Production-limitation cross-ref | No flags. The optional subtle push-in is flagged as default-skip if Soul drifts; safer to hold locked. |
| Comp library inspiration | [Anchor A — primary] sustained-talking-head close with persistent caption pattern. [Anchor B] integrated-into-routine CTA register (no salesy push, no urgency — the soft close converts on credibility, not pressure). [Anchor D — STRUCTURE ONLY] identity-restoration close pattern ("reclaim X" → "feel like myself again" is the cleanest version of that audience-call). NOT borrowing Anchor C/E's higher-pressure closes. |
04Visual continuity audit
| Audit category | Status | Notes |
|---|---|---|
| VH2 — location continuity | PASS | Kitchen-table setting sustained all 6 beats. The one B-roll cutaway (Beat 3b, 4 seconds — bedroom mechanism overlay) is the EXPLICIT mechanism visual; the post-cutaway close-up cut at t=0:52 lands cleanly back in the kitchen frame. |
| VH2 — lighting continuity | PASS | Warm window-left 5500K natural daylight sustained beats 1, 2, 3a, 3c, 3d, 4, 5, 6. Beat 3b cutaway intentionally palette-contrasts (dim blue-grey 3am bedroom) — the contrast IS the mechanism visual signal; clean return to warm on cut to close-up at t=0:52. |
| VH2 — avatar / wardrobe continuity | PASS | Single Soul Character locked across all 8 Soul renders (Beats 1, 2, 3a, 3c, 3d, 4, 5, 6 — even though split into 2 × 10s sub-renders per beat for Soul drift mitigation, the SOUL_ID + PROMPT stays identical). Sage-green cardigan + cream blouse + walnut table sustained. |
| VH2 — eyeline continuity | PASS | Direct-to-camera eyeline throughout — solo dynamic + at-home register requires sustained viewer eye-contact (per dynamics/solo.md §3 "Intimacy level: medium-high. Speaker is addressing the listener as one specific person"). |
| VS1 — B-roll cuing density | PASS — INTENTIONALLY UNDER-CUT | 1 B-roll insert across 2:00 (the 4s mechanism cutaway at Beat 3) = 1 per 120s. UNDER the 1/30s default floor — flagged but intentional per Field 21 + Tier 1 Fork 3 + Anchor A's 4-min talking-head proof. Confirmed intentional via Field 21 directive ("podcast-cut, minimal cuts") — NOT a refinement. |
| VS3 — comp library citation per beat | PASS | All 6 beats cite at least one anchor; Beats 3 + 5 also log original-direction rationales for specific sub-cues per VS3b (Beat 3 close-up cut + Beat 5 damaging-admission). |
| VS3b — per-B-roll citation | PASS (1 cue) | Only 1 B-roll cue in entire storyboard (Beat 3b mechanism cutaway). Cite + rationale logged: technique-borrowed from Anchor E, register adapted (painterly watercolor, NOT 3D Pixar). |
| VS4 — transition style consistency | PASS | Soft cuts throughout, ONE cross-dissolve (Beat 3a → 3b) and ONE hard cut (Beat 3b → 3c). Both belong to the same beat (Beat 3 — the mechanism reveal); the dissolve telegraphs entry-into-biology, the hard cut delivers the "your body isn't broken" reframe. Pattern is internally coherent within the script's one earned hard-pivot beat. |
| VS5 — mute-test asset survives 5s | PASS | Hook beat (Beat 1): speaker face + on-screen text-stagger survives full 18s (well past 5s). |
05Mute-test verification
Hook beat (0:00–0:18) — PASS via Fork 4 Strategy 1 (Speaker face + on-screen text).
- 70yo Natural-Health Authority face direct-to-camera (no movement, no zoom, sustained wide-medium framing) + on-screen text kinetic stagger of the three hook specificity beats: "3 in the morning." / "Heart slamming." / "Sense of doom." — lands all three captions across t=1.5s to t=12.5s.
- Mute-test scroller in first 5s sees: warm-lit kitchen + 70yo woman direct-eye-contact + "3 in the morning." caption fully visible by t=1.5s + "Heart slamming." caption fully visible by t=4.7s.
- VH1 ✓ / VS2 ✓ (≤4 words per caption) / VS5 ✓ (asset survives full 18s) / VS3 ✓ (Anchor A cornerstone + Anchor D structure cited).
Hook beat verbatim line on screen:
- "3 in the morning." (t=1.5s–4.5s)
- "Heart slamming." (t=4.7s–8.5s)
- "Sense of doom." (t=8.7s–12.5s)
Bonus mute-recapture beats:
- t=0:19–0:24 — Failed Solutions Stack stagger ("Melatonin. / Magnesium. / Weighted blanket. / Trazodone.") — a scroller landing at 20s sees the visual stack reinforce the failed-solutions list.
- t=0:48–0:52 — mechanism cutaway (painterly sleeping-woman + glucose curve) — visual scroll-recapture for mid-script landers.
- t=0:52–0:57 — close-up CUT to speaker face on the "your body isn't broken" reframe beat — visual SHIFT alone (wide-medium to close-up) re-engages a sound-off viewer.
- t=1:16–1:21 — "Fake keys jamming the locks." center italic — body-metaphor caption is its own mute-test asset.
- t=1:28–1:45 — formula ingredient stagger + brand-reveal sequence — strong mute-test asset for any sound-off viewer landing on the close.
- t=1:55–2:02 — "Said she felt like herself again." italic + persistent "trykindled.com" URL — soft CTA close passes mute test.
Script has FIVE distinct mute-recapture entry points across the 2:00 runtime (hook, failed-solutions, mechanism cutaway + close-up cut, body-metaphor, CTA arc) — structural strength for a deliberately under-cut talking-head.
06Production-limitation cross-ref
Per production_limitations.md heuristic: total non-template entries currently <3 (Phase 1 empty-doc state), so YS6 cross-check silently skips. No flag fires.
Three informational notes (NOT flags):
- Soul drift on sustained renders >8s. Per Soul Cinema known behavior. Workaround applied: each Soul beat split into 2 × ~10s sub-renders, with a natural blink-cycle at the composite seam. Composite in post.
- Soul rendering of on-screen text unreliable. Per
avatar_tool_prompt_syntax.md. ALL on-screen text in this storyboard is post-production overlay, NEVER prompted into Soul. Spec is unambiguous per beat. - Mechanism B-roll cutaway routed separately to Higgsfield Marketing Studio (image-to-video with painterly lifestyle preset). Soul does not produce abstract mechanism visualization well; Marketing Studio handles the soft watercolor + glucose-curve overlay cleanly.
07Tooling prompt index
| Beat | Tool | Prompts | Sub-render count |
|---|---|---|---|
| 1 (Hook) | Soul Cinema | 1 prompt block | 2 × 9s sub-renders |
| 2 (Validation) | Soul Cinema | 1 prompt block | 2 × 10s sub-renders |
| 3a (Mechanism explanation) | Soul Cinema | 1 prompt block | 1 × 10s render |
| 3b (Mechanism cutaway) | Marketing Studio (image-to-video) | 1 prompt block | 1 × 4s render |
| 3c (Close-up reframe) | Soul Cinema | 1 prompt block | 1 × 5s render |
| 3d (Return to wide-medium) | Soul Cinema | 1 prompt block | 1 × 8s render |
| 4 (Loop 2 close + cortisol) | Soul Cinema | 1 prompt block | 2 × 10s sub-renders |
| 5 (Unfair Advantage + product reveal) | Soul Cinema | 1 prompt block | 2 × 10s sub-renders |
| 6 (CTA close) | Soul Cinema | 1 prompt block | 1 × 16s sustained (or 2 × 8s with composite) |
| 6 post-card | Marketing Studio | 1 prompt block | 1 × 2s closing card |
Totals: ~10-12 Soul Cinema sub-renders + 2 Higgsfield Marketing Studio image-to-video renders + post-production text overlay layer.
08Pipeline update block
Storyboard produced: 2026-06-03 — C2 3AM Heart Slam (Laurie persona) via /storyboard-yapper Concept ID: C2 Session: kindled-batch-5 Production Target: Higgsfield Soul Cinema (kindled_speaker_1_1_natural_health_authority_v1 Soul Character) + Higgsfield Marketing Studio (1 mechanism cutaway B-roll + 1 closing card) Visual Style: Lived-in kitchen-table intimacy — warm daylight, walnut + cream + sage palette, sustained wide-medium with single close-up cut on the reframe beat Comp Library Anchors cited: 1142646044918657 (resilia Amish-grandmother, CORNERSTONE-VIDEO) / 1874728453166829 (happymammoth kitchen-confessional, video) / 759146623860120 (resilia oregano authority, video) / 1465022601769793 (trysoluma 3am-cortisol HPA, static — structure only) / 887515617080742 (tryjevi cortisol musical, video — technique only) Mute-test: PASS (5 mute-recapture entry points across 2:00) Production-limitation flags: 0 active; 3 informational notes (Soul drift workaround, text-in-post, Marketing Studio routing for cutaway) HARD rules audit: VH1 PASS / VH2 PASS / VH3 PASS / VH4 PASS SOFT rules audit: VS1 PASS-intentionally-under-cut / VS2 PASS / VS3 PASS / VS3b PASS / VS4 PASS / VS5 PASS