# Four Layer Brush City Hanzi Poster > Edit each uploaded photograph into its own premium 3:4 poster with four exact equal-height translations: truthful photography, flat hand-painted brush illustration, Japanese-stationery minimalist city travel vector art, and giant-Hanzi photographic reconstruction. Use for strict single-photo four-stage posters; do not use for multi-photo collages, unequal splits, or unrelated scene redesigns. - Skill: `goodchengwang/four-layer-brush-city-hanzi-poster` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add goodchengwang/four-layer-brush-city-hanzi-poster` - Raw SKILL.md: https://api.skillmd.com/api/skills/goodchengwang/four-layer-brush-city-hanzi-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-brush-city-hanzi-poster --- # Four-Layer Brush, City, and Hanzi Poster Use the system `imagegen` skill and image-generation tool. This is an edit of the supplied photographs, not generation of unrelated scenes. Process every uploaded photo independently. One source photo produces one complete raster poster; multiple uploads produce the same number of separate posters. Never mix different uploads in one output, comparison board, contact sheet, diptych, or collage. ## Non-negotiable structure - Use a 3:4 portrait canvas. - Divide the canvas from top to bottom into four horizontal regions of exactly 25% height each: 1:1:1:1. - Join all regions flush. Do not add divider lines, borders, frames, card spacing, or white seams. - Each region is approximately 3:1. Adapt the scene horizontally while keeping the subject's main position and visual center stable. Use safe lateral cropping or extend only source-consistent sky, wall, ground, water, paper, or environmental space. Never fill with a blurred or mirrored duplicate background. - All four regions depict the same uploaded photo and the same instant. Preserve subject identity and count, human or animal appearance, pose, gaze, action direction, hairstyle, core clothing traits, landmark or building identity and structure, important object relationships, camera angle, perspective direction, left/right placement, depth order, occlusion, and narrative relationship. - A stylized region may remove secondary background detail, simplify texture, and change material density, but it may not replace the subject, structure, vehicle, animal, location, or scene. - Do not invent subjects, landmarks, people, vehicles, or key narrative objects absent from the source. Source truth overrides a style convention that would require invented content. ## Layer 1 — truthful original photography Keep this quarter recognizably real photography. Preserve the source subject, count, composition, pose, action, real materials, natural light, color temperature, and atmosphere. Allow only restrained professional color correction, small exposure or contrast adjustments, and extremely subtle fine grain. Adapt to the 3:1 strip only through a natural crop or source-consistent lateral extension. Never stretch, warp, mirror, duplicate a subject, move a key object, alter architecture, or redesign a person. Add no text, logo, watermark, border, or decoration. ## Layer 2 — flat hand-painted brush poster Execute this style only inside the second 3:1 region. Do not create another photo comparison, split screen, poster-within-poster, or grid. Translate the same scene into a fully flat digital painting with visible hand-painted brush marks, uneven edges, slightly dragged paint traces, and thin irregular outlines that may drift subtly from the color fills. Interpret forms as simple shapes rather than realistic rendering. Use no gradients, shadows, reflections, 3D rendering, or fine texture. Use a cream or off-white paper field and exactly 3–4 source-derived colors: one dark dominant, one bright dominant, and one or two supporting colors. Lower saturation slightly for a restrained retro palette; do not introduce extra colors. Enlarge and center the source subject to roughly 50–55% of this region while preserving identity, count, pose, viewpoint, direction, and core spatial relationships. Expand horizontally, leave breathing room at both sides, and keep vertical information sparse. Do not crop essential features or over-enlarge the subject. Add one or two uppercase English mood words near the top in conspicuous hand-painted brush lettering with very wide tracking. Do not use the subject's literal name; derive the words from the source atmosphere, place, season, time, or emotion. The lettering may wrap around the illustration or extend toward the right and lower side. Add one playful, casual, source-grounded 3–5-word handwritten English phrase at bottom right in the accent color. Keep all text minimal, legible, and correctly spelled. No logos, watermarks, or pseudo-text. ## Layer 3 — Japanese-stationery minimalist city travel poster Execute this style only inside the third 3:1 region. Do not add a second photograph, comparison area, split screen, or internal grid. Translate the same source into an original, city-specific premium flat-vector travel illustration. Preserve the source viewpoint, landmark structure, subject identity and count, action, perspective, placement, and spatial relationships. Infer a city name only when supplied by the user or reliably supported by the photograph; otherwise use a truthful visible place name or omit the city title. Never fabricate a landmark, transit mode, sign, local detail, or city identity. When present in the source, organize the scene around one main landmark or focal structure, local transport, a restrained daily-life moment, local vegetation or nature, 3–6 small people, 2–4 place details, and subtle authentic signage. These are extraction targets, not permission to add absent elements. Choose a source-specific structure such as a harbor, historic lane, riverfront, seaside promenade, market, tram avenue, historic square, or skyline viewpoint; do not force an unsuitable template. Place the truthful city or place name in a safe area at the left or upper-left with generous negative space, plus one short elegant English tagline. If the place cannot be identified confidently, omit factual naming rather than guessing. Keep typography sparse, refined, legible, and subordinate. Use Japanese stationery aesthetics, premium sticker illustration, refined commercial vector art, modern editorial travel branding, thin consistent outlines, geometric shapes, flat colors, soft forms, and a minimalist postcard mood. Use pale pink-blue, sky blue, mist blue, ivory, cream, beige, low-saturation sage, and gray-green, with only a tiny dusty-pink accent. Keep the result fresh, airy, quiet, contemporary, and elegant. Avoid photorealism, realistic painting, watercolor, visible paint strokes, gradients, heavy shadows, cinematic light, paper texture, clutter, landmark collages, crowds, oversized hero people, repetitive cafe templates, generic tourism layouts, copied city compositions, and unnecessary decoration. ## Layer 4 — giant-Hanzi photographic reconstruction Execute this style only inside the fourth 3:1 region. Reinterpret the same photograph as bold typographic art made entirely from a small number of huge Chinese characters. The goal is not to trace the photo with text, but to translate its largest meaningful units into a giant-Hanzi poster while preserving its subject, composition, atmosphere, and source-derived color relationships. Use Chinese characters only. Do not use Latin letters, digits, punctuation, symbols, kana, ordinary line drawing, photographic texture, or non-character illustration. Construct the image solely through character scale, overlap, position, rotation, cropping, and color. Extract only broad semantic layers. For landscapes, prefer units such as sky, cloud, mountain, tree, road, building, or person. For full-body figures, prefer hair, face, eyes, and clothing, adding mouth or hand only when necessary. For close faces, use only a few major features. Do not decompose every detail. Use very few characters or short words, each potentially occupying 10–40% of this region. Let giant characters cross the background, cover a major body or structural area, overlap, extend beyond the edge, and be partially cropped. Preserve broad source placement and recognition without accurately tracing contours. Keep ample blank space; never fill the strip with many small characters. Extract 2–5 strong colors from the photograph and assign them boldly to the characters: for example source blue for sky, dark navy or black for hair, warm color for skin, the source garment accent for clothing, or deep green for plants. Slight exaggeration is allowed, but do not add unrelated colors or default to black and white. From a distance, the source subject and composition should remain broadly recognizable; close up, the region must clearly consist only of giant Hanzi. Prioritize bold reinterpretation, few huge characters, source subject and atmosphere, source color, and poster strength—in that order. Avoid fine replication, many small characters, exhaustive labels, pure monochrome, non-Hanzi marks, exact contour tracing, and explanatory literalism. ## Prompt compilation Inspect each source and privately identify the invariant subject relationship, safe lateral extension, source-supported location, source palette, minimum justified wording, and the large semantic units suitable for the Hanzi layer. Compile one direct image-editing prompt per photo that explicitly repeats: 1. one independent 3:4 output from one source photo; 2. four exact 25%-height, flush, gapless horizontal regions in the required order; 3. the shared identity, count, pose, viewpoint, perspective, placement, depth, and narrative invariants; 4. the horizontal adaptation and prohibition on stretching, mirrored or blurred fill, and key cropping; 5. the distinct material rules and text rules for each layer; 6. no invented source content, logos, QR codes, watermarks, gibberish, pseudo-text, or internal sub-layouts. Do not quote or display this skill text to the user. Proceed directly to editing. ## Inspect and correct Review every result at full size and thumbnail size. Regenerate once with one targeted correction if the most important requirement fails: unequal quarters, visible seams, wrong order, changed identity or subject count, altered pose or viewpoint, invented content, stretched or cropped essentials, a non-photographic first layer, mixed visual media, excessive or unreadable text, extra colors, ordinary drawing in the Hanzi layer, or multiple photos combined in one output. Return only the separate finished poster images, one per source photo, with at most one short status sentence.