| name | travel-memory-sticker-card |
|---|---|
| description | Transform a user-uploaded photograph into one horizontal collectible memory card with a large unframed marker-drawn illustration, six integrated journaling-sticker motifs, three scene-derived keywords, tactile paper tooth, and deliberately clumsy oversized flat color fields. Use when the user asks to P图, edit, redraw, stylize, or turn a travel, street, landscape, lifestyle, portrait, or pet photo into a minimalist postcard-like card or sticker-card in this visual language. Preserve recognition through one same-medium identification anchor and optionally one exact place-defining landmark text shown once; never retain an unintended photorealistic patch, watercolor wash, signature, or watermark. |
Travel Memory Sticker Card
Turn one user photo into one finished bitmap memory card. Use the image generation/editing tool; do not simulate the result with filters or code.
Workflow
- Inspect the source image at full useful detail.
- Identify the scene structure, emotional center, dominant spatial gesture, and one identification anchor.
- Decide whether one visible source text is a true landmark text anchor. Record its exact spelling and original placement; otherwise select none.
- Select six source-derived sticker motifs and exactly three concise scene-derived keywords before writing the prompt.
- Build the card using the fixed layout and art direction below. Read references/style-guide.md for scene-specific adaptation.
- Call the image editing tool with the user photo as the factual source. When available, use a previously approved card only as a layout/style reference, never as subject matter.
- Inspect the result. Regenerate once when the scene is not recognizable, the image looks photographic or polished, the anchor uses a different material, the card is crowded, sticker count is wrong, required text is misspelled, or unwanted marks appear. When landmark lettering alone is wrong, make one targeted text-correction edit instead of rebuilding the image.
- Show the best image and briefly name the identification anchor, landmark text if any, dominant quiet field, sticker motifs, and three keywords.
Fixed card system
- Use one 3:2 horizontal canvas on warm off-white uncoated paper.
- Keep a generous outer safe margin on all four sides, approximately 4–5% of the canvas width. Let the warm paper remain visibly continuous around the whole composition.
- Build the left 66–68% as one vertical unit: a large near-square main illustration above and a shallow keyword footer directly below it. Keep this whole unit slightly inset inside the outer paper margin.
- Treat the left illustration as an unframed rectangular image field laid directly on the card paper. Use a subtly rough, nearly square outer edge with no stroke, keyline, decorative border, white mat, inner card, rounded frame, drop shadow, sticker outline, or raised-paper effect.
- Do not place the illustration and keywords inside another enclosing rectangle. The card's outer paper margin is the only surrounding boundary.
- Center exactly three keywords beneath the main illustration, never beneath or inside the sticker column. Separate them with small centered dots:
[keyword 1] · [keyword 2] · [keyword 3]. - Place exactly six separate die-cut sticker motifs in the right 30–32%, using the full composition height beside the left image-plus-footer unit. Build a clear size hierarchy of two or three larger motifs plus smaller supporting motifs; stagger them with generous breathing room instead of making six equal catalog items.
- Give every sticker a thick irregular warm-white hand-cut border and a subtle flat paper shadow.
- Add exactly three short scene-derived keywords as a quiet centered typographic footer immediately below the left main illustration. Keep them small but readable, unboxed, and visually secondary.
- Use no writing area, postal markings, address lines, title, stamp box, subtitle, caption, or date. Allow only the three keywords and one optional landmark text anchor.
- Make the result read as one designed collectible card, not two pages or a busy sticker sheet. Preserve a relaxed handmade scatter rather than a rigid product grid.
Identification anchor
Use recognition, not photorealism.
- Choose one object or compact feature that makes the source identifiable: a roofline, umbrella, vehicle silhouette, window band, clothing color, branch fork, sign shape, shoreline, or pose.
- Preserve its silhouette, proportion, placement, relationship, and signature colors accurately.
- Render it in the same broad felt-tip marker medium as the entire card.
- Allow roughly 1.5–2 times the surrounding detail density, but never photographic texture, pixels, lens effects, or polished realism.
- Keep the anchor compact. When the main subject is large, preserve only its most distinctive structural cues.
- Do not let extra detail spread from the anchor into the rest of the scene.
Only retain a real photographic fragment when the user explicitly requests it or when identity-bearing content is essential, such as a face, handwriting, ticket, or irreplaceable material detail. Limit it to 5–10% of the artwork and integrate its edge softly.
Landmark text anchor
Preserve meaningful place recognition, not advertising clutter.
- Keep zero or one source-visible proper name, station label, distinctive storefront name, vehicle destination, or iconic product word only when it materially identifies the place or remembered object.
- Reject generic category or descriptive wording—such as ordinary menu items, prices, directions, and product types—unless the user says that wording itself carries the memory.
- Copy the text exactly from the source, including capitalization, script, and spacing. Never invent or autocomplete missing letters.
- Preserve its approximate position, scale relationship, orientation, and color role.
- Render the lettering in the same coarse felt-tip marker medium as the card; do not paste photographic signage.
- Allow a landmark brand or logo only when it is visibly present in the user's photo and functions as the scene's landmark. Do not add a brand absent from the source or turn the card into a promotional composition.
- Abstract every other advertisement, price, instruction, license plate, timestamp, and incidental text into unreadable color blocks.
- Show the landmark text exactly once across the entire card, normally in the main illustration. When the same sign becomes a sticker, keep its silhouette, colors, and layout but remove or abstract its letters.
- Quote exact landmark text verbatim in the image-generation prompt. Validate spelling at normal viewing size.
Art direction
- Rebuild the entire scene from 5–8 broad matte color families.
- Make the first read 3–5 genuinely oversized contiguous shapes. The dominant field and main subject masses must remain legible at thumbnail size.
- Make fewer, larger forms whenever uncertain. Do not subdivide a broad field merely to imitate photographic texture.
- Compress dense stone, leaves, windows, roof tiles, crowds, snow clumps, and street clutter into one to three lumpy masses per region. Never construct them from pebble-like dots, confetti, tiny repeated strokes, or mosaic fragments.
- Keep secondary internal shapes sparse and substantially larger than the paper grain. Concentrate extra structure only in the identification anchor.
- Favor blunt, naive, heavy, slightly misproportioned forms.
- Use broad opaque felt-tip or chisel-tip marker fills on lightly textured paper. Preserve visible marker character: directional fill bands, occasional overlapping passes, slightly uneven ink density, blunt stroke endings, dry-nib gaps, and soft bleed into paper tooth.
- Keep marker texture subordinate to the large shapes. Streaks must read as broad hand-filled passes, never as many tiny hatch marks.
- Keep edges slightly wobbly, blunt, and hand-drawn. Allow mild overrun and imperfect registration between adjoining color fields.
- Do not use transparent watercolor washes, wet-on-wet diffusion, watery blooms, pooling pigment, soft painted gradients, feathery watercolor edges, or layered translucent glazing.
- Expand one quiet field—sky, snow, water, wall, sand, court, road, or ground—when the source supports it.
- Compress foliage into a few lumpy masses; buildings into planes; vehicles into boxes and bands; people and animals into compact faceless silhouettes.
- Retain only one or two source-derived accent colors. Use near-black sparingly.
- Keep the mood observational, spacious, humane, slow, and lightly nostalgic.
- Remove or abstract all logos, advertisements, license plates, timestamps, and incidental readable text except the selected landmark text anchors. Always remove watermarks and signatures.
Sticker selection
Select six motifs only from meaningful visible source elements. Aim for:
- one main-subject sticker;
- one paired or grouped variation when supported;
- one environmental form such as foliage, snow, water, cloud, or rock;
- one structural fragment such as roofline, awning, window, railing, or color stripe;
- one functional object such as stool, cooler, sign, pole, bag, or vehicle detail;
- one small atmospheric or scale element.
Vary scale and rotation slightly. Keep the stickers coarse and in the same opaque felt-tip marker medium as the main image. Do not invent unrelated icons or make polished clip art. If a sticker repeats the landmark sign, remove its lettering so the exact text appears only once on the card.
Build each sticker from one to four chunky contiguous color masses. Do not reproduce foliage, stone, architecture, or texture with many tiny pieces.
Keywords
- Choose exactly three short words or compact phrases grounded in visible scene traits, distinctive objects, spatial qualities, weather, light, or the remembered mood created by those details.
- Prefer specific words such as “红坡道”, “海风”, or “石墙” over generic words such as “旅行”, “美好”, or “治愈”.
- Match the user's language; default to concise Chinese when the user writes in Chinese.
- Quote the three chosen keywords verbatim in the image-generation prompt.
- Set them once in a restrained footer centered directly beneath the left main illustration, separated by centered dots. Keep ample spacing, no label such as “KEYWORDS”, no boxes, no sticker outlines, and no repeated copy elsewhere.
Prompt skeleton
Replace every bracketed field:
Create one 3:2 horizontal collectible memory card from the supplied photo on warm off-white textured paper. Keep a continuous generous outer safe margin on all four sides, approximately 4–5% of the canvas width. Build the left 66–68% as one unit: a large near-square unframed illustration field above, with a shallow exposed-paper footer directly below it. The illustration lies flat on the card paper with a subtly rough rectangular edge: no stroke, keyline, decorative border, white mat, enclosing inner card, rounded frame, drop shadow, sticker outline, or raised-paper effect. Center exactly these three scene-derived keywords once on the exposed paper below the illustration, separated by centered dots: [keyword 1] · [keyword 2] · [keyword 3]. Never place the keywords inside the illustration, inside a border, or under the sticker column. Place exactly six source-derived die-cut journaling stickers in the right 30–32%, using the full height beside the left image-plus-footer unit. Arrange two or three larger hero stickers with smaller supporting stickers in a loose staggered cluster; give them generous irregular breathing room and avoid an equal-sized catalog grid. Preserve [identification anchor] through accurate silhouette, proportion, placement, and signature colors, but draw it in the same broad opaque felt-tip marker medium as everything else—no photographic patch. Preserve exactly one occurrence of this source-visible landmark text in its approximate original position in the main illustration: [verbatim landmark text, or NONE]. If a related sign appears as a sticker, remove its letters. Rebuild the scene so its first read is only 3–5 genuinely oversized contiguous color shapes using [palette and dominant quiet field]. Merge [dense source regions] into one to three lumpy masses per region; no pebble dots, confetti, mosaic fragments, tiny repeated strokes, or photo-like subdivision. Fill shapes with broad chisel-tip marker passes showing restrained directional bands, overlaps, uneven ink density, blunt endings, and occasional dry-nib gaps on paper tooth. Simplify [source elements] into naive, heavy, slightly misproportioned shapes with wobbly hand-drawn edges. Build every sticker from one to four chunky marker-filled color masses. Include these six stickers: [motifs]. Add no keyword label, caption, subtitle, date, or other readable text. No watercolor wash, transparent glazing, wet blooms, soft watercolor edges, writing area, postal marks, watermark, signature, photorealism, glossy 3D, smooth gradients, anime, polished vector art, or unrelated objects.
Quality checks
Approve only when all are true:
- The original scene remains identifiable at a glance.
- The whole artwork uses one consistent coarse medium; no region looks pasted from a photo.
- The identification anchor is recognizable through structure and color, not realistic texture.
- The first read is 3–5 oversized contiguous shapes and negative space, not detail.
- Dense regions are merged into a few broad masses; no area dissolves into pebble dots, confetti, tiny repeated marks, or mosaic fragments.
- The left illustration clearly dominates and the right area contains exactly six separated stickers.
- The left illustration has no frame, outline, white mat, inner-card container, rounded border, shadow, or sticker edge. Its subtly rough rectangular boundary meets the exposed card paper directly.
- The three keywords are centered directly beneath the left main illustration and do not enter the right sticker column.
- The keywords sit on exposed card paper outside the illustration, never inside a border or enclosing panel.
- A continuous warm-paper border remains visible on all four sides; the composition has approximately 4–5% outer breathing room and no important element crowds an edge.
- Sticker sizing has clear hierarchy and feels open and naturally scattered, never cramped or arranged as a rigid equal-sized catalog grid.
- Sticker borders are irregular and tactile; motifs are source-derived and stylistically consistent.
- Broad opaque marker passes are visibly present across the illustration and stickers; no area reads as transparent watercolor, wet wash, or softly blended paint.
- Exactly three specific scene-derived keywords appear once as a quiet footer beneath the left main illustration, separated by centered dots, with no label or decorative boxes.
- The selected landmark text, if any, appears exactly once, is spelled exactly, remains subordinate to the scene, and uses the same coarse medium.
- A sticker that repeats the landmark sign contains no readable duplicate lettering.
- No other readable advertisement, logo, license plate, timestamp, watermark, signature, or accidental text remains.
Delivery
Show the finished image. Report that built-in image editing was used, summarize the anchor and sticker choices, and provide the saved path when available.