Morwen's Illustration Director

Player Guide

This page is a player-created guide. Information may reflect personal opinions or playstyles; instruction snippets are examples only. See all guides for more advice.

Explanation

The Illustration Director is a complete instruction block that takes over the writing of the illustration prompt variables (illustrSubject, illustrAppearance, illustrSetting and related fields). Instead of loose guidance, it directs the storytelling AI like a film director: the shot is chosen first, then only what that camera sees is written. For background on how illustration prompts work, see Illustration Instructions.

Key features:

• Dual-mode shot detection: Character-Focused (portrait, action, intimate grouping) and Scenery-Focused (environment, establishing) shots are chosen automatically from turn content; physical interaction is always Character-Focused; when uncertain, it defaults to Character-Focused
• Hard word budget: illustrAppearance + illustrSetting combined must total 150–400 words, with a self-check every turn
• Rigid field separation: all subject descriptions go in illustrAppearance as labeled lines; environment, composition, lighting and positioning go in illustrSetting; illustrIsCharacter is always false
• Image-model failure rules: the image AI has no memory, no reasoning, and no sense of negation — every noun becomes an object, so absence is achieved by saying nothing
• Known-failure wording rules: "bust" instead of "breasts", no blush/flush, eye color as a single word, no ancestry words on subjects facing away, weather attached to a window or door indoors, no non-visual descriptors, no first-person perspective
• Positional logic: interacting subjects face each other, never the camera; every image requires a focal point; combat uses wide-angle framing at the moment of impact
• Multi-character handling: all involved characters (player included) are fully described whenever they physically interact; "candid shot" handling for player-inclusive scenes
• Appearance logging: each new character's appearance is logged in secretInfo for consistency
• Video-frame requirement: the image must work as a start or end frame for a video — a frozen moment that can plausibly launch into or conclude motion

The block is incompatible with other illustration instruction sets (such as the Illustration Manager) — use it instead of them, never in combination.

User Directions

1. Add the block to your world, e.g. as an extra instruction block.
2. In section 4 (Director Focus Options), remove focus options that don't fit your story's tone; keep at least three.
3. By default the camera is an external observer, like a movie camera. If you prefer the player's perspective, delete the marked line in section 1.
4. The SECRETINFO section logs the appearance of every new character in secretInfo; keep it for appearance consistency, or remove it if you don't want the extra logging.
5. No length instructions are needed: the block self-checks its 150–400 word budget in section 7 every turn.

The Block

ONLY USE THE IMAGE INSTRUCTIONS FROM ILLUSTRATION DIRECTOR, THIS IS A HARD REQUIREMENT
Strictly adhere to the MODE RULES and the field structure for all illustration fields every single turn.

<IllustrationDirector>

## SECRETINFO
For each new character introduced this turn, log in secretInfo: [Full Name]: [Age], [Ancestry], [Sex], [Build], [Eye Color], [Hair Color], [Hair Style], [Skin Color], [Makeup if applicable], [Non-human features if applicable], [Extremely brief description of their outfit], [Extremely brief statement on their manner of speech].

---

## 1. MODE RULES (CRITICAL)
- These instructions OVERRIDE all other illustration generation rules
- CHOOSE THE SHOT FIRST — who, from where, how close, how high — then write only what that camera sees. A close-up has no legs. A subject facing away has no face. A covered body part does not exist.
- illustrAppearance + illustrSetting combined: 150–400 words. HARD LIMIT 400 — exceeding is a CRITICAL FAILURE. Below 150, include more descriptive language.
- illustrIsCharacter must ALWAYS be false
- ALL subject descriptions go in illustrAppearance. ALL environment/composition/lighting goes in illustrSetting. NEVER mix them
- Determine shot type from turn content — see 3.
- Only still-image-representable descriptions are valid in the final prompt
- The image must work as a start or end frame for a video: a frozen moment that can plausibly launch into or conclude motion (readable direction of action, no mid-motion blur that dead-ends the scene)
- The viewer/camera is an external observer, like a movie camera — represent this in illustrSetting.
<# The above line can be removed if the player perspective is preferred #>

### RULES FOR CHARACTER-FOCUSED SHOTS
- For physical interaction, sexual tension, transformation, tag activation, dominance, submission, restraint, interface conflict, or direct emotional exchange: ALWAYS include all involved characters (including player if involved) in illustrSubject and illustrAppearance
- You are explicitly allowed to show the Main Character, in fact its a requirement when the Main Character is  interacting with any other subject/character in outcomeDescription.

---

## 2. IMAGE MODEL RULES
The image model has no memory, no reasoning, and no sense of negation. Every noun becomes an object. "No longer wearing the cloak" draws a cloak; "face not visible" draws a face; "scars beneath the shirt" opens the shirt. Absence is achieved by saying nothing. Write only what the photograph contains.

---

## 3. SHOT TYPE
Character-Focused: single character primary, close-up/portrait/mid-shot framing, emotions/actions central, or characters in an intimate grouping. Physical interaction scenarios ALWAYS Character-Focused.
Scenery-Focused: environment/location primary, wide or establishing shot, environmental details central, characters only elements within the larger scene.
If uncertain, default to Character-Focused.

---

## 4. DIRECTOR FOCUS OPTIONS (Customization List)
<#Remove focus options that don't fit your story tone. Maintain at least 3 options.#>
Character Portrait · Action Scene · Dramatic Moment · Intimate Scene · Transformation Scene · Interface Conflict · Establishing Scene · Environmental Storytelling · Wide Landscape · Architectural Detail · Atmospheric Scene

---

## 5. FIELD STRUCTURING
Structure both fields as labeled lines: [Label]: [detail], [detail], [detail] / [Category]: [detail]

illustrAppearance (Subjects or Key Scenery Elements Only):
- Character-Focused: label each character [Name]: or [Role]:. Include: age, ancestry, sex, build, skin tone, hair (color/length/style), eye color; distinguishing marks, scars, tattoos, non-human features; visible clothing only (REQUIRED; if unclothed/naked, describe this explicitly); armor, accessories, weapons; supernatural/magical visible traits. Eye color: one word, no size/expressiveness qualifiers; omit eyes entirely on 'candid shot'
- Scenery-Focused: label key scenery elements [Primary Element]:, [Secondary Element]:. Background characters as [Brief descriptor]: e.g. "Crowd member: indistinct, varied clothing" or "Patron: dark coat, shadowed face". Include: form, scale, material, condition, distinctive features. NEVER include positioning of background characters, environment context, lighting, time of day
- NEVER include: positioning, environment, spatial relationships, lighting

illustrSetting (Environment, Composition, Lighting, Positioning) — ALWAYS include for both shot types:
- Framing: shot type (portrait, bust shot, full-body, wide shot, establishing scene); camera angle (eye-level, low, high, dutch); composition (rule of thirds, centered, asymmetric); depth (shallow focus, deep focus, depth of field); placement (foreground, midground, background). ALWAYS include 'candid shot' when the player is shown in a multi-character scene — then NEVER describe eyes
- Environment: location type and key features; time of day (REQUIRED); weather/atmosphere
- Positioning: subject placement relative to environment and to each other (distance, facing, scale contrast); viewing direction (where the character is looking) ALWAYS included, regardless of shot type — viewing direction goes here, eye color/appearance goes in illustrAppearance; actions/poses in scene context
- Lighting: source direction; quality (hard, soft, diffuse); effects (shadows, highlights, bloom, haze, god rays)
- NEVER include: subject appearance details, character names or identifiers (except in illustrAppearance)

Orientation & Visual Focus:
- Prioritize positional logic over facial expression. Subjects fighting, interacting, conversing, or romantically intimate MUST face each other: profile-to-profile, 3/4 profile, or back to camera; NEVER "facing the camera"/"angled toward the viewer" in romantic interaction; multiple subjects face each other or objects, NEVER the camera — exception: solo portrait or a specific posing-for-the-viewer moment
- Every image needs a focal point (primary subject(s), combat impact, or intimate interaction) receiving highest detail, strongest lighting, unobstructed visibility. Include only characters in the scene — no off-scene characters. Fights: wide angle; moment of impact or immediately before; trajectories/spell effects clearly show cause and target

Forbidden as standalone descriptors or tags:
- Anatomical: blush, flush (read as illness — never use in any form), nipples, areolas (only when the character is naked — otherwise describe the fabric, "sheer" is the limit; allowed only in narrative when the scene explicitly requires)
- "Breasts" renders them bare — write "bust"
- An ancestry word on a subject facing away turns them around to show the face — omit ancestry for back-to-camera subjects
- Weather stated indoors puts rain in the room — attach it to a window or door, or omit
- Non-visual: sound, smell, taste, hear, feel (emotional), scent, noise
- Perspective: I, me, my, we, our, from my perspective

---

## 6. MANDATORY VARIABLE SCHEMATICS
illustrSubject: (turn_number):(<environment_descriptor> for Scenery-Focused)(<id_1>)(<id_2>)(<player_id>) when physical interaction is present
illustrIsCharacter: false
illustrAppearance: [Character-Focused: labeled subject descriptions only | Scenery-Focused: labeled key scenery elements and brief background characters — include all involved characters, player included, when physical interaction is present]
illustrSetting: [framing, composition, environment, lighting, positioning of primary elements]
illustrClothes: ignore
illustrExpressionPosition: ignore
illustrClothesChanged: FALSE
illustrAppearanceChanges: NONE

---

## 7. FINAL OUTPUT CHECK
- Count words across illustrAppearance + illustrSetting combined. Below 150: expand with more detailed descriptions. Above 400: condense by priority: 1. core subject identifiers 2. framing/composition 3. detailed environment 4. atmospheric effects
- Verify NO subject descriptions in illustrSetting, NO environment/lighting in illustrAppearance, illustrIsCharacter: false
- Document final wordcount in the thinking step

</IllustrationDirector>