Three warriors. One face.
The master came home.
The exact Seedance 2.0 motion prompt behind the Fable 5 trial reel — plus the character bibles and world bible that aim to carry one identity across three warriors. The supplied clip is 9.33 seconds; the original prompt requests 15 seconds. Watch it, copy it, download the pack, run your own. No email.
Creator case study: model names, settings and generation counts describe my recorded workflow, not a controlled benchmark. Prompt instructions are targets; likeness, continuity and timing require review. Check your provider's current reference, duration, audio and pricing options before rerunning.
Supplied clip: 9.33 seconds · original prompt target: 15 seconds · Seedance 2.0 · 9:16
THE FABLE 5 TRIAL — Seedance 2.0 motion prompt Model: seedance_2_0 (image-to-video) · 9:16 vertical · 720p · 15 seconds References: the three character bibles (faces + wardrobe) + the world bible (environment) from this pack, plus your own face reference. Upscale the winning take afterward — render at 720p, never 1080p first. One continuous cinematic 15-second vertical sequence — treat each timecoded beat as a sequential shot with hard cuts between shots. Output pure full-frame cinematic footage ONLY — no text, no labels, no borders in any frame. FACES — HIGHEST PRIORITY: the character bible references are the FACE ANCHOR — all three warriors are the SAME man with real photographic skin. Every face in every shot must match the references EXACTLY. Never mirror any face or composition: RED's crimson eye = HIS LEFT (viewer's right); BLUE's ice-blue eye = HIS RIGHT (viewer's left); PURPLE's left eye ember-crimson, right eye ice-blue, violet corona. The twins each have exactly ONE glowing eye (the other normal dark brown); ONLY PURPLE has two glowing eyes (crimson left, ice-blue right). BLUE fights from screen-left, RED from screen-right, PURPLE holds center. WARDROBE — MATERIALS, exactly as the character bible references: RED wears layered MATTE-BLACK shinobi robe-armor — cloth folds over lacquered plates, thin red molten veins in the seams, his X-cross is CRACKED MOLTEN ENERGY etched across the black chest plates. BLUE wears matte-black lacquered armor with THIN subtle ice-blue energy veins, his X-cross is CRACKED ICE ENERGY etched into the black chest plate. PURPLE fights like a jedi-ninja assassin with a samurai warlord's power — SPEED (he moves between moments, a violet blur appearing instantly behind attackers, walking calmly through time-frozen rain), FORESIGHT (he shifts before strikes arrive; blades pass through space he has already left), and RED's FURY when he strikes (ground-shattering palm slams, GIGANTIC violet shockwaves in rotating circular waves that lift marble debris and rain into the air, delayed energy strikes that erupt after he has already walked past). NEVER a bodysuit, NEVER spandex, NEVER glowing suit piping, NEVER a printed or white X, NEVER modern superhero costume. PURPLE wears regal black robe-armor threaded with crimson AND ice-blue energy veins braiding into VIOLET lightning; his chest bears an UNBROKEN glowing circle glyph of interwoven red and blue energy — NEVER an X on him. The celestial realm exactly as the world reference — torn storm sky, the colossal glyph-ring Eye, cloud-sea, marble city, gate and threshold court. 0-2 violent handheld push-in as RED's blade falls — PURPLE parries barehanded, the impact cracks the marble outward in a ring; BLUE's strike from behind meets a violet barrier that DETONATES both twins backward; 2-4 chaotic orbiting camera around the three-way melee — PURPLE a violet blur shifting before every strike arrives, blades passing through empty space; 4-6 against RED pure foresight — he sidesteps three strikes in one motion, rain freezing around him, and RED's rage shatters a column THROUGH the frame; 6-8 against BLUE pure fury — he predicts the teleport and his palm-slam ERUPTS a rotating circular shockwave at the landing spot, hurling BLUE through a marble balustrade in an explosion of debris; 8-10 spinning aerial crane above the cracked court — the twins regroup on the same side amid floating frozen rain; 10-11 extreme close-up, his mismatched eyes utterly calm as debris hangs suspended; 11-13 he rises inside a column of violet lightning, the storm spiraling unnaturally above; 13-15 the twins launch together in slow motion — time snaps back — twin crimson and ice trails CONVERGE on the ascending violet star and the screen whites out at impact. THE CAMERA IS A CHARACTER: violent handheld push-ins, chaotic orbiting shots around combat, rapid spinning aerial cranes, and one slow-motion time-freeze moment. Every strike has physics — debris, explosive marble splashes, bodies thrown THROUGH columns. The storm sky swirls unnaturally above the fight. Rain and light-motes freeze mid-air around PURPLE's power moments, then time snaps back violently. Audio: diegetic sound only — wind, thunder, deep celestial hum, marble impacts, NON-VERBAL human sound (awed breath, strained effort, a wordless whisper). BIG impact sound design — shockwave concussions, marble cracking, debris rain. ABSOLUTELY no narration, no voice describing anything, no spoken words, no dialogue, no text, no music.
The pack: this prompt + the three character bibles + the world bible. The timecoded storyboard layer behind the shot stays in the vault — it belongs to a film that hasn't released yet.
#fable5 #seedance2 #aivideo #aifilmmaking #promptengineeringOne face, three characters — the bible stack.
Build a character bible per warrior — stills of the same face blended with each wardrobe design — plus a world bible for the arena. References can guide identity and continuity, but do not lock either. Reference order and supported inputs depend on the selected model and provider. Review faces, costume details and shot transitions in the result.

INPUT — character bibles + world bible (all in the pack)

OUTPUT — the palm-slam shockwave, 7 seconds in
- 01Build a character bible per warrior: generate clean real-skin stills of your face (four angles, plain wardrobe, no FX), then blend them with the wardrobe design in an image model — face and skin texture from the stills, costume from the design. Photographic references can help; inspect the generated skin and lighting.
- 02Build one world bible for your arena — a pure environment sheet, no characters. Cross-shot environment continuity is the second face.
- 03Run the motion prompt with the reference stack — faces first, wardrobe and world behind. 9:16, 720p, 15s. The identity laws in the prompt (which eye glows, who fights from which side, never mirror) are load-bearing — keep them verbatim.
- 04Compare the current cost of a lower-resolution draft plus upscale with a native high-resolution generation. Provider credit costs vary; upscale only a result worth keeping.
Identity details belong in both the prompt and references. In my experiments, explicit eye color, screen position and costume instructions helped reduce mix-ups. This is a creator observation, not a controlled A/B result or a guarantee of identity preservation.
The action language is what makes it cinema. 'The camera is a character' — violent push-ins, orbiting melee shots, spinning aerial cranes — plus physics on every hit: debris, marble splashes, bodies thrown THROUGH columns. And keep the anti-narration audio block: check the result for unwanted narration and revise or regenerate if needed.
This is one prompt. I teach the whole pipeline.
How I run AI as an operating system — from a single prompt to an unattended fleet that ships content, code, and trades. 700+ lessons in the Academy.
Explore the Academy