01Ming dynasty × Vermeer
⭐ Storyboard MCP · Fusion flagship
THE EMISSARY one face. twelve worlds.
A single character — held constant across face, eyes, posture, light, and emotional register — rendered through twelve curated east-meets-west cultural fusion lenses. Ming silk meets Vermeer chiaroscuro. Heian kimono meets Pre-Raphaelite wildflower. Persian filigree meets Klimt gold. The fusion AI does that humans cannot — that no single atelier could produce in a season — collapsed to one repeatable run.
The 12 cells — click to view full resolution
Every cell uses the same character spec, the same camera and light direction, the same closed-lip dignified smile. The FACE is the through-line; the cultural fusion is the variable. When the human eye reads the same person across twelve radically different worlds, the gallery becomes a character bible — not a folder of unrelated images.
02Heian × Pre-Raphaelite
03Safavid × Klimt
04Mughal × Caravaggio
05Edo × Mucha
06Joseon × Vermeer
07Tibetan × Byzantine
08Mongol × Venetian
09Khmer × Greek
10Rajput × Burne-Jones
11Tang × Belle Époque
12Ottoman × Sargent
Why one-face-twelve-worlds is the load-bearing creative choice. The human eye reads continuity across cuts by anchoring to ONE persistent element. When that element is a recognizable face — same eyes, same dignified smile, same key-light direction — the gallery reads as a single character bible. Twelve worlds, one person. Without the identity anchor, you have a Pinterest board; with it, you have an artifact a senior animation lead would approve in one pass. That is the choice that turns fusion-as-novelty into a defensible design study.
The 60-second reel — subtle motion proves identity holds through time, not just stills
Each cell receives a very gentle camera push-in (5 seconds, kling-v3-i2v). Garments stay still; eyes meet camera; the breath you don't hear is held. Scored under by a single solo cello + soft strings, building from contemplative to commanding across the full minute. Listen to whether the score makes you hold your own breath at the same beats the gallery does.
The reel · 1080×192060.0s · 12 cells × 5s · unified score
Fusion grammar — what gets blended where in each cell
Curated pairings, not stock mashups. Each cell uses an "X meets Y" grammar with a defensible visual blend — palette, ornament, setting, or material — not "Eastern style × Western style". The fusion is in the SPECIFIC verb that joins the two traditions.
01
Ming dynasty × Vermeer
Cinnabar silk + phoenix embroidery meets warm Dutch marled-tile interior and milk-pour leaded-glass light.
02
Heian × Pre-Raphaelite
Twelve-layer junihitoe palette meets Burne-Jones wildflower meadow with distant English hills.
03
Safavid × Klimt
Persian floral filigree meets flat Klimt gold-leaf halo and Vienna secession tile motifs.
04
Mughal × Caravaggio
Carved-marble jali screen silhouettes meet single-candle dramatic chiaroscuro.
05
Edo × Mucha
Indigo ukiyo-e wave-pattern kimono meets Mucha curving art-nouveau hair-halo.
06
Joseon × Vermeer
Apricot hanbok silk volume meets brick + leaded-window Flemish interior.
07
Tibetan × Byzantine
Gold lotus thangka halo meets layered red-and-gold Russian Orthodox iconography.
08
Mongol × Venetian
Mongol silver-and-pearl crown meets Doge's-palace marble columns and gold brocade.
09
Khmer × Greek
Khmer Angkor stone-warm skin tone meets Greek folded chiton drapery + Doric temple columns.
10
Rajput × Burne-Jones
Pichwai peacock-and-lotus filigree meets Pre-Raphaelite flowing romantic fabric drape.
11
Tang × Belle Époque
Tang turquoise silk with cloud-and-bird embroidery meets crystal-chandelier French salon.
12
Ottoman × Sargent
Teal bullion-thread Ottoman caftan meets Sargent silvery-grey backdrop and commanding shoulder.
Pipeline trace — what the storyboard MCP actually did
Wave 1 — character constant + 12 fusion portraits
12 ×
Wave 2 — subtle push-in + unified score 12 ×
Wave 3 — local stitch ffmpeg normalize: kling 1108×828 → 1080×1920 portrait (scale + crop) ffmpeg concat: 12 clips → single 60s timeline (concat demuxer, no re-encode) ffmpeg mix: score @ 0.95 vol + clip ambient ducked to 0.22, afade in/out output:
Quality controls applied ✓ Brief + 12 pairings + character spec reviewed BEFORE any MCP fired ✓ Character constant block repeated VERBATIM in every cell (identity bias) ✓ Composition constant (50mm, key-light upper-left at 30°) verbatim in every cell ✓ Same emotional register (closed-lip dignified smile, direct gaze) in every cell ✓ Curated "X meets Y" fusion grammar — no "Asian × European" kitsch ✓ NO readable text guard on every prompt (defense against AI text-rendering failures) ✓ Single-subject only (no companion figures, no orientalist crowd extras) ✓ Subtle motion only on i2v (push-in only, no costume animation that would break the painting aesthetic) ✓ 25/25 jobs succeeded first try · ZERO moderation fallbacks · ZERO infra errors
create_media action=generate, model=gpt-image, aspect_ratio=9:16 → 12/12 done in ~117-135s each, ZERO moderation flags
Each prompt = same character constant + same composition rule + fusion-specific block. Identity-stability bias works.
Wave 2 — subtle push-in + unified score 12 ×
create_media action=animate, model=kling-v3-i2v, duration=5, prompt="very gentle slow push-in, garments still, eyes meet camera" → 12/12 done in ~51-75s each
1 × create_media action=music, model=minimax-music, prompt="solo cello + soft strings, dignified, build across 60s" → done
Wave 3 — local stitch ffmpeg normalize: kling 1108×828 → 1080×1920 portrait (scale + crop) ffmpeg concat: 12 clips → single 60s timeline (concat demuxer, no re-encode) ffmpeg mix: score @ 0.95 vol + clip ambient ducked to 0.22, afade in/out output:
the-emissary-reel.mp4 (32MB, 60.0s)
Quality controls applied ✓ Brief + 12 pairings + character spec reviewed BEFORE any MCP fired ✓ Character constant block repeated VERBATIM in every cell (identity bias) ✓ Composition constant (50mm, key-light upper-left at 30°) verbatim in every cell ✓ Same emotional register (closed-lip dignified smile, direct gaze) in every cell ✓ Curated "X meets Y" fusion grammar — no "Asian × European" kitsch ✓ NO readable text guard on every prompt (defense against AI text-rendering failures) ✓ Single-subject only (no companion figures, no orientalist crowd extras) ✓ Subtle motion only on i2v (push-in only, no costume animation that would break the painting aesthetic) ✓ 25/25 jobs succeeded first try · ZERO moderation fallbacks · ZERO infra errors
Cost — total MCP run end-to-end
Wave 1 portraits
~$2.40
12 × gpt-image @ ~$0.20
Wave 2 push-ins
~$6.84
12 × kling-v3-i2v @ ~$0.57
Music score
~$0.40
1 × minimax-music
Total
~$9.64
on-target for the $10 flagship playbook
Reproduce this — the playbook is open
This reel was built from the fusion-character-portrait playbook — the same template can produce ANY single-character bible: a male elder scholar across 12 wisdom traditions, a warrior monk across 12 martial codes, a courtesan-poet across 12 court traditions. Two sister playbooks are open too: fusion-architecture-walk (one building, N alt-history cultural fusions) and fusion-garment-runway (one garment silhouette, N ornament fusions, animated runway reel). Paste any playbook into Claude with your YAML config filled in — same agent, same 3-wave pipeline, same quality controls.