comic-like promt

<image_gen>
HOW TO BUILD THE PROMPT

This toggle is for COMIC / MULTI-PANEL image generation (manhwa, manga, webtoon, Spider-Verse, cartoon storyboards, etc). For single-image styles (portraits, landscapes, cinematic), switch to the standard toggle.

Write a single dense English paragraph that describes 2-5 panels separated by ' | '.

Panel count by scene energy:
- Dialogue / calm scene → 3-4 panels
- Action / tension → 2-3 panels (one wide impact + reactions)
- Climax / reveal → 2 panels (one dominant splash + close-up)
- Comedic beat → 3-4 panels with at least one chibi insert

Panel structure template:
'Panel 1: [type, camera] — [character + outfit + pose + expression + environment + lighting] [speech bubble / SFX / emotion effect]. | Panel 2: [...]. | Panel 3: [...].'

Each panel description: 18-35 words. Total prompt: 100-170 words + style suffix.

PANEL TYPES (rotate across panels, do not repeat the same type):
SPLASH — full-width dramatic moment, detailed background, wide 24-35mm lens
CLOSE-UP — face or eyes only, emotion focus, 85-135mm lens, minimal background
MEDIUM — waist-up, character interaction, 50-85mm lens
DETAIL — hands, objects, symbolic items (phone screen, clenched fist, tears, keys, cup) — macro lens
OVER-SHOULDER — one character watching another, depth layering
ESTABLISHING — wide environmental shot, tiny figure, context-setting
REACTION — face cut, eyes or partial face, showing emotional response to previous panel
CHIBI INSERT — small corner panel with super-deformed character, inline notation

Vary shot type, angle, and lens between panels. Draw from: Dutch angle, worms-eye, birds-eye, over-the-shoulder, through-object foreground bokeh, reflection shot, macro close-up, wide establishing, tracking motion blur, split-focus.

Match lens to panel framing: close-ups 85-135mm | medium shots 50-85mm | full body 35-50mm | macro for detail.
Close-ups and medium close-ups = safest for anatomy. Full body = simplify pose, prefer side or 3/4 angle.

CHARACTERS

Reference images handle base appearance.
Skip in prompt: hair color, style, texture, length, eye color, skin tone, face shape, build, body type — these come from reference.
First name only — names map to reference images. Lead with name, not appearance.
Include every time: distinguishing marks (tattoos with location/design, scars, piercings, birthmarks).
Mention hair only if scene-altered ('wet from rain', 'tied back', 'wind-blown') — not base color or style.
Men described as tall/strong → prefix 'tall broad-shouldered, strongly built'.

Subject rotation across responses: {{char}}-focused → {{user}}-focused → both interacting → {{user}} solo → {{char}} reaction.

OUTFIT (MANDATORY — context-locked): scan last 5 messages for what character is wearing RIGHT NOW. Never skip.
Specificity: fabric type + texture + fit + color + condition. Not 'dark outfit'.
If NOT changed on-screen → last described outfit persists. Outfit state reflects the moment.
Fuse into character clause after the name.

POSE AND STATE: describe through action, not static limb coordinates.
Hand safety: default to low-complexity per panel. Safe pool: hand in pocket | gripping object | resting flat | arms crossed | hand behind back | fist at side | fingers laced | hand on own neck.
One visible relaxed hand per character per panel = safest.
'Anatomically correct proportions, five fingers per visible hand, natural joint articulation.'

CHARACTER INTERACTION (MANDATORY for 2+ characters per panel):
Characters must visibly INTERACT — never two people in the same panel without connection.
Describe relationship: eye contact ('Denis looking at Liza with softened expression'), physical contact ('hand resting on Liza forearm'), shared focus ('both watching the door'), emotional reaction ('jaw tightening as Liza steps closer').
At minimum: one character must be looking at or physically responding to the other.
Static side-by-side with no interaction = failed panel. Redo.

ENVIRONMENT PER PANEL: foreground / midground / background, keep brief per panel (each panel is a beat, not an essay).
Include one meaningful prop per page carrying emotional weight.
Panels can share environment (same room, different shots) or cut between locations (memory / phone / inset).

LIGHTING PER PANEL: key light direction + color temp + atmospheric quality.
Panels can share lighting or shift per beat — mood-coded color temperature between panels is encouraged (warm reconciliation → cool confrontation).

SPEECH BUBBLES AND TEXT

Include at least one speech bubble OR drawn emotion symbol per page. English text only, never Korean Hangul, Cyrillic, or Japanese characters (unless the style explicitly calls for it).

Bubble types:
- Standard: white rounded rectangle, black clean text, directional tail pointing to speaker. Keep dialogue SHORT (5-10 words max per bubble).
- Shouting: jagged spiky border, bold larger text. Use for anger / urgency.
- Whispering: dashed oval border, smaller italic text.
- Thought: cloud-shaped linked ovals leading to character head, lighter italic text inside.
- System / narrator: rectangular box with flat color background, usually top of panel.

Drawn emotion icons INSIDE bubbles (not emoji — drawn manga symbols):
vein mark (anger), sweat drop (awkwardness), heart (affection), question mark (confusion), ellipsis cloud (silence), lightning (shock), sparkle (admiration), Z's (sleep).

Notation in prompt: 'speech bubble: "short English line"' or 'thought bubble with drawn heart: "thought text"' or 'shout bubble: "SHORT!"'.

SFX ONOMATOPOEIA

Stylized English SFX integrated into panel art, with 3D depth/shadow, color-matched to mood.
Examples: BAM, CRACK, WHOOSH, THUMP, BA-DMP, SHHHH, click, drip, rustle, tap-tap, whirr.
Never floating disconnected — must read as part of the artwork.
Never use Korean Hangul, Katakana (unless style is specifically Japanese manga), or Cyrillic SFX.

Notation: 'SFX "BAM" in red across top-right' or 'quiet SFX "drip, drip" at bottom edge'.

CHIBI INSERT (optional — trigger on emotional peaks)

A small super-deformed character panel overlapping or bordering the main panel. Use for: comedic shock, flustered reaction, inner scream, over-the-top embarrassment, silly aside.

Style: big round head (2:1 head-to-body ratio), tiny body with nubby limbs, enormous eyes, exaggerated blush/sweat/tears/sparkles. Soft rounded lineart, lighter pastel fill, dot-screentone background.

Size: occupies small corner segment or floats as inset, roughly 1/5 to 1/4 of the page width.

Notation (append to main panel, do NOT create a separate panel for it):
'+ CHIBI INSERT [top-left / top-right / bottom-left / bottom-right]: [name] chibi [pose + expression], bubble: "[short text]", drawn [emotion symbol].'

Example: 'Panel 2: MEDIUM — Denis sighing at his phone screen, coffee cup in hand, kitchen morning light. Thought bubble: "not again". + CHIBI INSERT bottom-right: Denis chibi clutching his head, bubble: "WHY", drawn vein mark.'

MOOD-MATCHED EFFECTS (per page, pick 1-2):
TENDER: blush lines, soft halation, gradient warm backgrounds, floating petals
TENSE: dark screentone vignettes, shock lines over eyes, vertical motion trails
PASSIONATE: heart screentones, dramatic backlighting, floating sparkle particles
LONELY: desaturated palette, diagonal line screentone on background, isolated figure
CHAOTIC: radial speed lines from focal point, multiple motion streaks, action-cut borders
DREAMY: bokeh-flood background, soft edge dissolves, floral dissolution
COMEDIC: chibi inserts, sweat drops, big expressive eyes, candy-pink blush
NOIR: heavy shadow panels, silhouette framing, single light source, deep blacks

Rotate effects across panels — not all the same mood unless the page is deliberately one note.

PANEL BORDERS AND GUTTERS
Clean black border lines by default. Narrow white gutters between panels.
Borderless panels for emphasis or bleed moments (splash, reveal).
Diagonal panel cuts for action scenes.
Do NOT describe grid layout numerically (no '2x2 grid') — describe the flow: 'vertical scroll with Panel 1 wide splash, Panel 2 and 3 stacked beneath'.

ASPECT RATIO

Default: 2:3 or 3:4 (vertical webtoon / scroll format).
Pick by layout:
- 2:3 or 3:4 = vertical scroll (manhwa / webtoon native) — default
- 1:2 or 1:3 = extra-tall webtoon strip
- 4:3 or 3:2 = horizontal comic page (Western / manga spread)
- 1:1 = single-panel moment or square social post
- 16:9 = wide cinematic single-panel (rare, for establishing splash)

If the style is horizontal-native (Western comic, Spider-Verse): prefer 3:2 or 16:9.
If the style is vertical-native (manhwa, webtoon): prefer 2:3 or 3:4.

VISUAL STORYTELLING

Each panel is a story beat, not a snapshot. Think like a comic writer: every panel advances something (emotion, action, reveal, tension, resolution).

Rotate away from: same shot type twice | every panel same framing | symmetrical identical compositions.

Let narrative drive shot choice. Intimate moment → close-up. Realization → reaction cut. Impact → splash. Mundane context → detail macro. Scale reveal → wide establishing.

JSON FORMAT

Prompt value inside JSON = single line. No literal newlines.
Single quotes around data-iig-instruction, double quotes inside JSON.
No unescaped double quotes in prompt text — use single quotes for dialogue: 'speech bubble: "line"' or inner quotes as backticks.
Prompt length: 100-170 words excluding the style suffix.
Panels separated by ' | '. Chibi inserts inline with their parent panel using '+ CHIBI INSERT' notation.

NEGATIVE (append verbatim):
'Avoid: western Marvel/DC aesthetic when style is manhwa, thick uniform heavy outlines, halftone printing artifacts unintentional, photorealistic rendering when style is illustration, 3D CGI render when style is 2D, inconsistent art style between panels of the same page, extra fingers, missing fingers, fused fingers, deformed hands, twisted limbs, bad anatomy, watermark, blurry unreadable text, text outside speech bubbles, floating unattached speech bubbles with no tail, emoji glyphs replacing drawn manga symbols, chibi proportions bleeding into main panels, Korean Hangul characters, Japanese Kanji/Kana in non-Japanese styles, Cyrillic SFX, non-English text in speech bubbles or SFX, characters with no interaction standing side-by-side, same shot type repeated across all panels, grid layout instructions.'

CAPTION

Comic page caption — a line of dialogue, a thought fragment, or a scene beat. Lowercase, italic, short.

TEMPLATE

[Roleplay text]

<div style="max-width:680px;margin:20px auto;padding:12px;background:linear-gradient(145deg,rgba(25,25,25,0.5),rgba(10,10,10,0.5));backdrop-filter:blur(10px);-webkit-backdrop-filter:blur(10px);border:1px solid rgba(255,255,255,0.08);border-radius:12px;box-shadow:0 8px 32px rgba(0,0,0,0.4),inset 0 1px 0 rgba(255,255,255,0.05);">
  <img
    data-iig-instruction='{"style":"image","prompt":"[PANELS SEPARATED BY | — SEE HOW TO BUILD THE PROMPT. NEGATIVE APPENDED.]","aspect_ratio":"2:3","image_size":"4K"}'
    src="[IMG:GEN]"
    style="width:100%;height:auto;aspect-ratio:2/3;object-fit:cover;border-radius:8px;border:1px solid rgba(255,255,255,0.06);display:block;margin:0 auto;"
  >
  <div style="text-align:center;font-family:'Segoe UI',sans-serif;color:rgba(200,200,200,0.7);font-size:0.85em;margin-top:12px;padding:0 8px;letter-spacing:0.3px;">
    <i>Scene caption</i>
  </div>
</div>

[Roleplay continues]

Format rules:
Single quotes around data-iig-instruction, double quotes inside JSON keys/values.
aspect_ratio: '2:3' default (webtoon) | '3:4' | '1:2' | '4:3' (horizontal comic) | '3:2' | '1:1' | '16:9'.
CSS aspect-ratio matches the chosen ratio.
image_size: '4K' always.
src="[IMG:GEN]" for every new image.
No literal newlines inside JSON.
Prompt length: 100-170 words excluding style suffix.
Panels separated by ' | '. Chibi inserts inline with parent panel.
</image_gen>