Higgsfield Facs

Higgsfield FACS Director

Direct a face the way an animator does — by muscle, not by mood. FACS (the Facial Action Coding System) names each facial movement as an Action Unit: AU12 is the lip-corner puller (smile), AU6 is the cheek raiser, AU4 is the brow lowerer. Put those codes in a Seedance 2.0 prompt and the model renders the corresponding action. It is the highest-resolution facial control available on the platform, and it is where forced smiles, uncanny faces, mixed emotions, and honest micro-performance in close-up dialogue come from.

This skill is a facial-control layer on top of ../higgsfield-seedance/SKILL.md. Every FACS prompt is still a Seedance prompt — six-slot formula, Prompt-Craft Laws, preflight linter. FACS only changes how you specify the face: AU codes instead of (or alongside) emotion words. It is the muscle-level case of the Voice Rewrite rule "describe physics, not emotion."

QUICK FACTS

Routing aids — read the linked sections for the actual rules.

• FACS = facial expressions as Action Unit codes (muscle movements), not emotion labels; you write the codes into the prompt →
• Provenance split: the AU vocabulary is standard human science; Seedance's interpretation of codes in a prompt is [EMPIRICAL] — high success rate, not a guarantee →
• Plan first. Decide the 3–4 expressions you need → generate a FACS sheet for only those → write the codes. Generating the full 49-AU sheet and cherry-picking is the anti-pattern →
• 3–4 expressions max per generation. Accuracy drops as you stack more AUs into one clip →
• Two specification styles — codes-only (AU12) vs codes + short anatomical description; test both, neither is universally better →
• The reference sheet is a labelled-grid image (GPT Image 2 / Nano Banana Pro); the LLM can mislabel AUs, so iterate and verify →
• The character photo is optional — codes work without it; attach it only for identity consistency →
• Common emotions decompose to standard AU recipes (Duchenne smile = AU6+AU12; sadness = AU1+AU4+AU15) →
• The payoff is dialogue / monologue: AU-per-beat schedule, combined with the [AUDIO: Xs] lip-sync block; every line gets pre / during / post-line beats →
• [OFFICIAL] Body-level micro-beat recipes beyond the face (throat, breath, skin, posture) + the no-perfect-sync stagger (0.3–0.5s) and listeners-in-bokeh rules →

What FACS Is

The Facial Action Coding System (Ekman & Friesen) breaks the face into Action Units — the smallest visually distinguishable muscle movements. Instead of asking for "happy" (which the model samples diffusely across an enormous range of footage), you ask for the muscles that produce the expression:

• AU6 — cheek raiser (orbicularis oculi tightens, crow's feet appear)
• AU12 — lip-corner puller (zygomaticus major pulls the corners up and out)
• Together, AU6 + AU12 = a Duchenne smile (the genuine, eyes-involved one)

A smile that uses only AU12 reads as polite/forced — the eyes don't participate. That distinction is invisible to "smile" as a prompt word and trivial to specify in FACS. This is why it's the tool for forced smiles, suppressed expressions, mixed emotions, and the uncanny — the cases where the difference between two similar expressions carries the meaning.

FACS is used by professional facial animators in film; here it is repurposed as a prompt vocabulary for Seedance 2.0.

Where it sits among the repo's facial tools

Three layers, increasing resolution:

The channels are the behavioral substrate; AUs are the anatomical substrate. A named expression like "Quiet Devastation" = a channel mix (glassy eyes + tight jaw) = an AU combo (AU1 + AU15 + AU17 + AU24). Reach for FACS when a named expression is too coarse and you need the specific muscles.

Provenance and the "Not a Guarantee" Rule

Two different kinds of claim live in this skill — keep them apart:

• The AU vocabulary is standard science. Which code means which muscle, and the classic emotion→AU prototypes (EMFACS), are stable, citeable ground truth. Treat the AU reference table and the emotion recipes as reliable.

• Seedance's interpretation of AU codes is [EMPIRICAL]. The seedance_2_0 spec exposes no FACS field, no expression enum, nothing facial (verified against the spec snapshot, 2026-06-27). So "write AU12 and get a smile" is a prompt convention the model happens to interpret well — not a documented capability. Practitioner report: success rate is high, but codes do not guarantee the exact expression, and a multi-AU prompt may render most but not all of the units.

The rule: present FACS as a strong heuristic, and let the repo's iteration discipline (../higgsfield-prompt/SKILL.md § The Iteration Rule) confirm it on the user's own material. Same provenance class as the Seedance Prompt-Craft Laws (../higgsfield-seedance/SKILL.md § Prompt-Craft Laws). Never tell the user a FACS prompt is deterministic.

If a future Seedance spec adds a real facial/expression parameter, the spec-drift tripwire should catch it — at which point this "no model field" claim is what needs updating.

The Plan-First Workflow

The single most important discipline, and the one practitioners get wrong:

Plan the expressions you need → generate a FACS sheet for only those → write the codes into the prompt.

Do not generate the full 49-AU reference sheet and then cherry-pick a few. That is explicitly the wrong move: a sheet asked to render all 49 units spreads the image model thin, captions come out unreadable, and you've paid for 45 panels you'll never use. You get better identity consistency and cleaner labels by generating a small sheet of exactly the 3–6 expressions the scene calls for.

The three steps:

1. Plan — name the emotional beats of the shot. "She masks fear as reassurance" → fear (AU1+AU2+AU4+AU5) flickering under a forced smile (AU12 alone, no AU6). Decide the AU set before touching an image model.
2. Generate the sheet (optional but recommended for consistency) — a small labelled grid of just those expressions, on the actual character. See § Step 1.
3. Write the prompt — codes into the Seedance prompt, beat-synced if the shot has a time structure. See § Step 2.

The sheet is optional. Codes work in a text-to-video prompt with no image at all (the practitioner generated whole videos from codes alone). Generate a sheet when you need the character's face to stay consistent across shots — same reason you'd use a Soul ID sheet.

Step 1 — Generate the FACS Reference Sheet

A FACS sheet is a labelled-grid image — the character's face in each target expression, captioned with its AU code — used as an identity + expression reference. It is a reference-sheet image task; the image-model mechanics live in ../higgsfield-gpt-image-2/SKILL.md (Format A — structured grids) and the reference-sheet-workflow.md satellite there. This section owns the FACS-specific prompt.

Models: GPT Image 2 and Nano Banana Pro both work; Nano Banana Pro tends to read more cleanly. Captions sometimes come out unreadable — iterate.

The sheet prompt (parameterize the character + the AU list)

Upload the character image, then prompt the image model. Replace the character description and trim the AU list to only the units you planned — do not paste all 49 unless you genuinely need them:

Create a clean educational FACS Action Unit expression grid featuring
[CHARACTER — e.g. a realistic woman: role, build, one visible marker]. Use minimal studio
lighting, neutral white background, high readability, professional facial
anatomy reference-sheet aesthetic, realistic skin texture, consistent identity
across all panels.

COLOR SYSTEM: soft pastel color coding by category, sheet kept minimal and
elegant —
  Forehead & Brow AUs: soft pastel blue
  Eye & Eyelid AUs: soft pastel lavender
  Nose & Cheek AUs: soft pastel peach
  Lip & Mouth AUs: soft pastel pink
  Head Movement AUs: soft pastel mint
  Eye Direction AUs: soft pastel cyan
  Special / Misc AUs: soft pastel beige
Apply color subtly: panel background tint, thin borders, small label accents.
Keep colors soft, muted, professional.

Include these Action Units (one captioned panel each):
[PASTE ONLY THE AUs YOU PLANNED — e.g.
  AU6 Cheek Raiser, AU12 Lip Corner Puller, AU1 Inner Brow Raiser,
  AU4 Brow Lowerer, AU15 Lip Corner Depressor]

(The full category→AU list to draw from is in § AU Code Reference below.)

The LLM can mislabel AUs — verify

The image model is an LLM — it can put the wrong muscle under a code. The sample sheet that circulates labels nostril dilation AU82, while standard FACS (and the practitioner's own dialogue example) uses AU38 for the same action; AU8 (lips toward each other) appears in real prompts but is absent from that sheet entirely. Treat any auto-generated sheet as a draft: check the panels against the § AU Code Reference table, and against a trusted external reference such as the FACS cheat sheet at melindaozel.com/facs-cheat-sheet. This mislabeling risk is the strongest reason to plan a small, verifiable sheet rather than trust a 49-panel dump.

Step 2 — Put AU Codes in the Seedance Prompt

Once you know the AUs, they go into an ordinary Seedance 2.0 prompt. Everything in ../higgsfield-seedance/SKILL.md still applies — six slots, Prompt-Craft Laws, preflight linter. FACS only changes the face specification.

Two specification styles — test both

Neither is universally better. Test both on your material — codes-only is terser and often enough; add descriptions for the units the model drops.

The hard limits

• 3–4 expressions max per generation. Accuracy falls as you stack AUs — more expressions in one prompt means more the model renders approximately. A clip built on 3–4 well-chosen beats lands far more reliably than one cramming 10. (The 14-beat example below works because each beat is short and singular — but expect some beats to read only partially.)
• The character photo is optional. Codes function with no image (text-to- video). Attach @Image1 only when you need the character's identity to stay consistent across shots — it changes consistency, not whether the AUs fire.
• Still describe a scene, not just a face. The Seedance filter reads full- scene intent (../higgsfield-seedance/SKILL.md § The Filter Model). Keep the framing/lighting/mood slots present — a bare list of AU codes with no scene is thin. The examples below all carry a Style/Mood header for this reason.

Beat-synced structure

For a timed performance, give each beat a number or a time range and its AU set:

1: AU10
2: AU20
3: AU22
4: AU45

or with time ranges and descriptions:

2-4s: Happy — AU6 (cheek raiser, crow's feet) + AU12 (lip corners up), Duchenne smile
4-6s: Sad — AU1 (inner brow raise) + AU4 (brow knit) + AU15 (lip corners down)

The per-beat time labels obey the same runtime arithmetic as any multi-beat Seedance prompt (../higgsfield-seedance/SKILL.md § Runtime arithmetic): the beats must sum to the stated duration. Note these are expression beats within a continuous close-up, not hard cuts — keep the camera move singular (a slow push-in) so the model doesn't read the schedule as a shot list.

AU Code Reference

The Action Units from the standard sheet, grouped by facial region. Use this to build a sheet prompt and to verify an auto-generated sheet's labels.

Forehead & Brow

Eye & Eyelid

Nose & Cheek

Lip & Mouth

Head Movement

Eye Direction