agentleFS
nigo-studionigo-studio/ai-film-skills/.claude/skills/character-consistency/SKILL.md

character-consistency

Skill15 starsChanged 3 months agoMITView on GitHub

Keep one character identical across every generated shot. Expand a short Character Brief into ultra-detailed generation prompts, build a three-panel character sheet (bust-up, full body front, full body rear) from a single close-up, and attach it as an identity reference on every generation. Use when the user wants the same character to stay consistent across multiple shots, scenes, or takes.

5 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions5 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep the user's character identical across every generated shot.

    line 8
  2. Distinctive:

    If the Brief came with a close-up image, skip the close-up generation — but still write out the identity blocks below from the Brief and …

    line 37
  3. Distinctive:

    Keep the sheet plain: Do not bake your film's color grade or style into it — a neutral studio sheet composites into any scene.

    line 95
  4. Distinctive:

    When delivering the sheet prompt, also hand the user the role line they will need afterward — it goes into every later prompt where the …

    line 100
  5. Distinctive:

    Every Brief field was expanded into specifics.

    line 112
  6. Distinctive:

    All three panels read as the same person in the same outfit.

    line 113

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Character Consistency

Keep the user's character identical across every generated shot. The method: expand a short Brief into ultra-detailed prompts, turn the prompts into a character sheet, and ride the sheet on every generation the character appears in.

The core job of this skill is expansion. Every one-line field in the Brief becomes a paragraph of specifics in the prompt. Vague prompts produce generic faces — and a generic face cannot stay consistent, because there is nothing distinctive for the next generation to hold on to. Detail is what consistency is made of.

What it does

Input: a filled Character Brief (below). A freeform description is fine too — map it into the Brief and show the result back to the user before writing any prompt.

Output: an ultra-detailed close-up prompt and a three-panel sheet prompt (bust-up / full body front / full body rear) for the user to run, plus the role line the user will carry into every later prompt where the character appears. This skill writes prompts only — the user runs the generations and picks the candidates.

Step 0. Fill the Character Brief

NAME:        (name or handle)
AGE / BUILD: (age impression, gender, build and height impression)
FACE:        (eyes, skin, identifying features — be specific)
HAIR:        (color, length, texture, how it parts and falls)
OUTFIT:      (top, bottom, shoes, accessories — item by item: garment, material,
             color, wear. The full-body panels are built from this text;
             anything not written here does not exist.)
CLOSE-UP:    (attach an existing close-up image if you have one;
             if not, the skill writes the close-up prompt from this Brief)
NOTES:       (optional: overall vibe, visual style — photoreal, anime, etc.)

Blank fields are allowed: fill them with conservative proposals and show the user the completed Brief. Ask a question only when two readings would produce different people — decide everything else and move on.

Step 1. Lock the close-up

If the Brief came with a close-up image, skip the close-up generation — but still write out the identity blocks below from the Brief and the image; Step 2 needs them.

Otherwise, expand the Brief into an ultra-detailed close-up prompt with this structure and hand it to the user:

Ultra-detailed portrait of <NAME>: <one-line summary from the Brief>.
Chest-up framing, front view, facing camera, plain cool grey background,
even soft studio light.

EYES (highest priority):
<iris color and texture, pupil, catchlights, eyelids, brow shape — expanded
from FACE. The eyes are where identity lives; spend the most words here.>

SKIN:
<tone, texture, lines, scars, stubble, marks — the details that make this
face this face and no other.>

HAIR:
<color, length, texture, how it parts and falls — expanded from HAIR.>

EXPRESSION:
<the character's default expression, precisely: what the mouth, brows, and
gaze do at rest.>

CLOTHING (visible in frame):
<the collar and shoulders of the OUTFIT as they appear chest-up.>

TECHNICAL:
<visual style from NOTES — photorealistic, anime, etc.> Sharp focus on the
eyes. High resolution.

Have the user generate several candidates with this prompt and pick one. The picked image is now the character — every later asset derives from it — so advise the user not to settle for a face that is merely acceptable: it should be one they would recognize in a crowd.

Step 2. Generate the three-panel sheet

Write the sheet prompt below and have the user run it with the close-up attached as the reference image — the output is one 16:9 sheet with three panels: bust-up front, full body front, full body rear. The IDENTITY block re-states everything — the close-up shows nothing below the chest, so the full-body panels are built entirely from the OUTFIT text.

Character reference sheet of the same person, 16:9 landscape format,
three panels side by side on a single seamless cool grey studio background,
identical grey tone across all three panels.
Left panel (about 45% of frame width): bust-up portrait framed from the chest up,
front view, facing camera, head height about one third of the panel height.
Center panel: full body standing straight, front view, arms relaxed at sides,
feet visible, figure filling about 90% of the panel height.
Right panel: full body standing straight, rear view, seen from directly behind,
heels visible, figure filling about 90% of the panel height.
Even studio lighting, consistent appearance, hairstyle and outfit across all three panels.

IDENTITY:
<the EYES / SKIN / HAIR blocks from Step 1, then the complete OUTFIT item by
item: garment, material, color, wear.>

Practical notes, each paid for in retakes:

  • Cool grey beats white. A white background clips highlights and destabilizes identity; a seamless cool grey keeps all three panels on the same footing.
  • Keep the sheet plain. Do not bake your film's color grade or style into it — a neutral studio sheet composites into any scene; a stylized one fights every scene it enters.
  • Panel ratios are probabilistic. Some runs squeeze the panels or drift toward equal thirds. Have the user generate a few candidates and pick the cleanest — do not fight a bad layout with prompt edits.
  • Removing gear the close-up shows (headphones, goggles, gloves): write the state that should exist, never a negation — bare head with natural hair only, bare empty hands.
  • A new outfit means a new sheet. Advise the user to make a second sheet for a wardrobe change rather than describing the change ad hoc.

When delivering the sheet prompt, also hand the user the role line they will need afterward — it goes into every later prompt where the character appears:

@Image 1 defines <NAME>'s facial features, hairstyle, and clothing.
Do not use the grey background or the panel layout.

How that reference is then written into a full video prompt — verbal reinforcement, several characters sharing a frame — is the companion skill, Video Prompt Director.

Self-check

  • The Brief is complete — especially OUTFIT, item by item — before any prompt is written.
  • Every Brief field was expanded into specifics; no field survived as a vague one-liner.
  • All three panels read as the same person in the same outfit; the sheet is plain grey with no scene style baked in.
  • The role line was delivered together with the sheet prompt.

What this skill does not do

  • It does not write the full video prompt — that is Video Prompt Director. This skill supplies the identity assets that prompt rides on.
  • It does not call any generation API. It writes the prompts; the user runs them.
  • It does not make generation deterministic. Sheets and shots are still probabilistic; the method turns "hope the face holds" into "pick the take where everything else is right, because the face already holds."

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.