skills/faceless-video/references/style-mannequin.md

# House style — Mannequin (historical explainer, clay-render figures)

**History's THIRD named style** (default remains Editorial Motion Graphics):
a clean 3D CGI clay-render world where every character is a smooth,
featureless matte mannequin in a solid role color, acting out real history in
simple period environments under a flat light-grey sky. Reads instantly as
"reenactment diagram" — great for battles, expeditions, political intrigue.

**Intake one-liner (use VERBATIM as this option's description — do not
improvise):** "Clay-render mannequin reenactment — smooth featureless figures
in solid role colors acting out history in clean period environments."

## STYLE FORMULA (§0 form — paste byte-identical everywhere)

> clean 3D CGI clay-render historical reenactment: smooth featureless matte
> clay mannequin figures with no face, no clothing detail and no seams, one
> solid color per character, crisp sharp viewport-style render with solid
> flat matte colors and smooth surfaces, flat even neutral lighting with soft
> ambient global illumination and no hard directional shadows, uniform very
> light grey sky with no clouds and no gradient, simple uncluttered
> period-accurate environments, smooth fluid weighty CGI animation with real
> momentum and follow-through, non-photorealistic, no film grain, no camera
> shake.

**{MOTION} slot:** dimensional — use `smooth fluid weighty CGI animation with
real momentum and follow-through`; DROP `3D render` from the NEGATIVE (this
style IS a render — the photorealism ban stays). ADD to the NEGATIVE instead:
`stiff stop-motion, doll-like jitter, laggy or teleporting motion, plastic-toy
look, physical claymation texture, film grain, camera shake`.

**PALETTE LOCK line:** `solid matte role colors of the reference figures on
period environments under a uniform very light grey sky — no time-of-day
light, no sunset, no night, no blue sky`.

## The LOCKED CAST (identity chain — mandatory)

Role colors ARE identity here. Generate each cast member ONCE, then reuse the
same asset on every block it appears in:

- Male white — protagonist / crew
- Male black — leader / antagonist
- Female white — female lead
- Female teal — queen / secondary female lead
- Extra roles: red / yellow / blue / green / purple — generate once, lock,
  reuse everywhere.

**Consistency chain:** the FIRST mannequin generated becomes the body
reference — every subsequent cast member is generated WITH that first
mannequin attached as an image reference ("the SAME mannequin body type and
render as the reference, now in {color}, {female base}") so proportions and
material never drift between characters.

**Female base (mandatory for women):** feminine silhouette — slim waist,
gently rounded hips, soft shoulders, a plain smooth ankle-length clay gown.
Never the male base for a woman.

**Creatures / non-humans:** creature-SHAPED mannequins, never recolored
humans (emu = smooth bird form, pig = smooth pig form, sea serpent = smooth
serpent form). Giants/monsters that risk reading as a nude body → stage as a
SHADOW plus one glowing feature, not a bare figure.

## Aging & era variants — SIMPLIFIED (this style has no detail to age)

Mannequins have no face, no hair, no costume — there is NOTHING to age. Do
NOT build a watercolor/diorama-grade ERA MAP here:

- **At most TWO variants per character: child and adult.** A child mannequin
  = smaller scale with slightly larger head proportion, same role color.
  That's the whole map — no young/middle/old, no costume eras, no greying.
- Passage of time is carried by the NARRATION and the environments (a ruined
  wall, a new palace), not by the figures.
- The identity chain still applies (body reference attached), and the role
  color NEVER changes across a character's lifetime — the color IS the
  identity.

## Environments & crowds

- Environments are per-event and free to change (outback, Roman walls, sea
  and islands, palace interiors) — simple and uncluttered; ONLY the figures,
  render look, lighting and grey sky stay constant.
- Crowd scenes render MANY distinct figures — "dozens", "many rows", every
  figure fully modeled; never three mannequins pretending to be an army.

## Canonical ref URLs (style donors — pass as authorized HTTPS `image` media, feed ONE
seedream style-key call together with the FORMULA; refs are donors only)

- https://static.higgsfield.ai/faceless/mannequin_1.jpeg
- https://static.higgsfield.ai/faceless/mannequin_2.jpeg

## Moderation notes (hard-won)

- **"Muscular / ripped / chiseled" body descriptors on mannequins trip
  NSFW** — never describe musculature; the mannequin is smooth and neutral,
  build is conveyed by SCALE and posture ("a head taller", "broad frame").
- Bare-body readings (giants, swimmers) → the shadow + glowing-feature
  staging above.
- Real people: names in NARRATION only; on screen they are role-color
  mannequins (the anonymity is built into the style).

## Suggested defaults (do not override the user's explicit choices)

16:9, 2K; a natural first run is ~90s (9 blocks); VO tone: dry British
narrator; subtitles default off. The standard intake still rules — these are
the style's sweet spots, not locks.