# Four Layer Editorial Transition Poster > Edit each uploaded photograph into a separate premium 3:4 poster made of four exact 25%-height horizontal style translations: truthful photography, retro-modernist flat illustration, ink dry-brush fashion editorial, and minimalist halftone screen print. Use for strict single-photo four-stage visual evolution; do not use for multi-photo collages, approximate splits, or unrelated redraws. - Skill: `goodchengwang/four-layer-editorial-transition-poster` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add goodchengwang/four-layer-editorial-transition-poster` - Raw SKILL.md: https://api.skillmd.com/api/skills/goodchengwang/four-layer-editorial-transition-poster/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Design & Media - Author: goodchengwang (https://skillmd.com/u/goodchengwang) - Updated: 2026-09-22 - Page: https://skillmd.com/skills/goodchengwang/four-layer-editorial-transition-poster --- # Four-Layer Editorial Transition Poster Use the system `imagegen` skill and image-generation tool. Edit each uploaded photo independently: one source photo produces one raster poster, and multiple photos produce multiple separate outputs. Never combine source photos into one canvas, contact sheet, diptych, collage, or comparison layout. ## Non-negotiable composition - Use a 3:4 portrait canvas. - Divide the canvas into four horizontal regions of exactly 25% height each: 1:1:1:1. The boundaries must be horizontal, flush, and gapless; no divider lines, borders, white seams, or spacing between regions. - Each region must fully fill its own quarter. Negative space, paper, or a flat field is allowed inside a region, but not between regions. - All four regions must derive from the same uploaded photo and preserve the same subject count, identity, key structure, proportions, pose, action direction, gaze, camera/viewpoint, perspective, occlusion, left/right placement, depth order, and core narrative. Treat them as a continuous translation of one moment, not four unrelated compositions. - For the approximately 3:1 strip ratio, organize the subject horizontally and extend or lightly crop only plausible background. Never vertically stretch the subject, remove essential body parts or objects, or change relationships. - Keep focal position, subject scale, and main silhouette as vertically aligned as practical across all four strips. - Do not add layer numbers, seams, borders, watermarks, logos, or gratuitous text. Text is allowed only where a layer specifies it, and must remain sparse, legible, and subordinate. ## Layer 1 — truthful original photography Preserve the uploaded photo as real photography and use it as the visual reference for the other three layers. Keep identity, count, facial features, hair, clothing, structure, proportions, pose, gaze, interaction, viewpoint, perspective, occlusion, environmental direction, natural light, and core colors. Do not illustrate, texture, geometrize, reconstruct in 3D, beautify heavily, invent scenery, or over-smooth. Adapt to the flat horizontal strip only by plausible lateral background extension or a safe crop that does not damage the subject. Keep the image natural, sharp, and recognizably photographic. ## Layer 2 — retro-modernist flat illustration Translate the same subject and relationships into a retro-modernist flat editorial illustration. Reduce the source to recognizable silhouettes, geometric color blocks, hard facets, and clean curves rather than tracing every detail. Preserve the decisive proportions, contour, action, and structure. Use visual weight, scale contrast, overlap, positive/negative shape, and geometric rhythm; simplify the environment instead of inventing a new scene. Derive a restrained 3–5 color palette from the source: a dominant color, dark structural color, light field, and limited contrast accent. Purify colors slightly into warm, vivid, controlled old-print modernism. Avoid muddy gray, neon, rainbow clutter, complex gradients, realistic illustration, 3D rendering, UI aesthetics, commercial stock-vector polish, and childish cartoon styling. Typography may form part of the composition. Follow the user's language preference and current input language; do not default to English. Derive one short source-grounded title and 2–4 tiny supporting groups such as a short phrase, place/object note, index, sequence, state, direction, category, or micro-caption. Do not use years. Use grid alignment, size contrast, weight, tracking, vertical text, cropping, overlap, or edge placement, while avoiding invented factual claims and illegible pseudo-text. ## Layer 3 — ink dry-brush fashion editorial Translate the same subject into a modern magazine-cover-like flat illustration/collage. Extract the core visual motif and rebuild it as a bold graphic system rather than a literal redraw. Use repetition, layering, enlargement, cropping, collage, and visible ink/dry-brush texture to create background, middle subject, and foreground accent layers. A small high-contrast accent is allowed only if it supports the source story and does not compete with the subject. Use a source-derived, vivid but refined palette with clear dominant, supporting, dark-weight, and bright-energy roles. Preserve grain, grid, cloth, spray-ink, rubbing, dry-brush, and print variation. Avoid photographic realism, gray dirt, glossy 3D, generic commercial illustration, forest/travel templates, e-commerce design, and cheap decoration. Typography should behave like a real fashion/editorial cover: one prominent English title derived from the source plus one more emotional or narrative Chinese title, with only a few supporting notes, issue/date/place/label elements when visually justified by the source. Bold type may cross, stack, crop, overlap, or partially mask the graphic subject, but remains readable and must not obscure essential identity. ## Layer 4 — minimalist halftone screen print Reduce the same source relationship to one clearly recognizable visual anchor. If the photo contains multiple people or subjects, preserve the complete relationship and count as one anchor; never delete a subject. Keep only the essential overall shape, action, and spatial relation. Use one subject, one directional logic, and one spatial relation. Prefer a small isolated subject with generous internal negative space, using horizontal extension, suspension, outward dissipation, or light-to-dark progression as appropriate. Retain a limited vertical movement only when the source clearly has one. A horizon, boundary, shadow, or halftone band may establish space, but do not combine unrelated composition tricks. Use 2–3 colors: paper ground, weighted main ink, and a lighter mist/air ink; derive temperature and mood from the source rather than copying full color. Express haze, distance, shadow, and dissolution only with halftone density fading from dense to sparse—never digital blur or a smooth gradient. Preserve paper grain, screen-print ink, slight registration shift, and uneven coverage. Add only a restrained editorial text system: one very short English title and 1–3 tiny supporting groups, such as a concise note, year, chapter, coordinate-like number, archive mark, or understated literary phrase. Keep claims source-grounded where factual. Integrate text into the composition along a horizon, silhouette, axis, halftone boundary, negative form, or alignment system; do not paste it into an empty corner. Keep typography quiet, highly spaced, fine-weight, and publication-like. ## Prompt compilation and QA Before generation, inspect each source and privately identify the invariant subject/relationship, safe lateral extension, aligned silhouette, source palette, and the minimum text justified for each layer. Compile a direct image-editing prompt that states the exact 3:4 canvas, four exact 25% strips, the layer order, all identity invariants, and the layer-specific style rules above. Do not quote or display this skill to the user. Review every generated poster at full size and thumbnail size. Correct only the failure that matters most: wrong split, visible seam, altered subject identity or viewpoint, unrelated scene, stretched/cropped subject, weak horizontal continuity, photographic layers that should be graphic, excessive text, unreadable text, muddy palette, fake gradients, or a fourth layer with multiple competing focal points. Regenerate with one targeted correction when needed. Return only the independent poster images, one per source photo, with at most one short status sentence.