
unknownPosters & Visuals
Have been working on a manga engine. It's a work in progress but you just need name, cha…
unknown prompt and image example
@Gdgtify
View source post ↗👁 164♥ 4★ quality 90
PROMPTTranslation
Have been working on a manga engine. It's a work in progress but you just need name, character ideas, things that fit your story to use a prompt like this on GPT.
2x2 grid, 16:9, do this for 4 famous ai inferred: char_seed ::= name + apparent_age + archetype + silhouette_anchor + color_key
- silhouette_anchor is the consistency lock: hair shape (wolf-cut twin tails,
stubborn ahoge, undercut braid) plus one accessory that never leaves
(headphone halo, cracked visor, bone hairpin) — identifiable in pure black at 64px
- color_key = 3 flats + 1 accent; the accent appears in the eyes and one prop, nowhere else
- eye construction locked and reused: iris shape, highlight count + position,
lash mass as a shape (never individual hairs), lower-lid line presence
- original character only — no existing franchise likeness, insignia, or trademarked design
input ::= char_seed + slice_beat + emotional_turn
ink_1 :: read the beat
- morning-rush / café-work / last-train / con-day / 3am convenience store
- one mundane object anchors the whole strip: a melting drink, a dying phone, a lost stamp card
- background fidelity is tiered: full-render establishing, tone-only mid panels, blank behind emotion close-ups
- one recurring background gag that rewards a reread
ink_2 :: read the turn
- the page exists for one emotional pivot; everything before it is setup, everything after is landing
- 4–6 panels; the turn panel takes 40% of page area
- tilted panel = agitation, bleed panel = scale, borderless panel = memory or interior monologue
- reading flow right-to-left in a z-path; gutters 4mm between beats, 2mm within one
tone :: screentone discipline
- 10% for skin shadow, 30% for clothing, 50% + gradient for mood; solid black for hair mass and dread
- hatching reserved for embarrassment and shock; speed lines for motion, focus lines for revelation
- sfx drawn as artwork inside the panel — deformed by perspective, never floating decals
- if ink can do it, tone doesn't
stage :: page layout
- b5 proportion, printer margins respected, one bleed panel maximum per page
- balloons placed along the reading path, tails aimed at the mouth, three balloon shapes maximum
- character consistency checked panel-to-panel against silhouette_anchor
bar :: an editor follows the flow with all text removed, the turn lands in the largest panel, and no tone exists that ink couldn't have done better
render_mode ::= pick one; it governs the entire pipeline downstream
- cel-2d: two shadow steps + one rim, hard terminators, line-weight hierarchy
- mono-manga: black/white + screentone (10/30/50%), zero grey paint
- painterly-splash: soft blending, occlusion, rendered fabric, illustration-thick
- toon-3d: cel shader over geometry, ink outline pass, real light rig
- pixel: fixed palette (16–32), declared grid (32/48/64px), no anti-alias on outline
#manga#comic#black and white#storyboard#anime style#grid layout
Related prompts

✨ View prompt♥ 56

✨ View prompt♥ 63

✨ View prompt♥ 84

✨ View prompt♥ 56

✨ View prompt♥ 30

✨ View prompt♥ 38

✨ View prompt♥ 55

✨ View prompt♥ 15

✨ View prompt♥ 26

✨ View prompt♥ 5

✨ View prompt♥ 27

✨ View prompt♥ 86