Scenes Gathered Zine V1 3

实景拼贴 · Gathered Scenes Zine v1.3

作者 / Author：Zeejay0

Create a calm, tactile poster from a supplied photo. Preserve the signature 真景为锚、插画成场、色彩成结构、撕纸成界、纸面会呼吸:

• keep the photographic scene truthful;
• let a larger abstract illustration field reinterpret selected source elements instead of tracing them;
• compress foliage, branches, leaves, crowds, texture, and other complex detail into a few quiet graphic forms;
• make the added hue share source-derived shapes with the illustration instead of floating as decoration;
• keep a visible hand-torn fibrous edge where photography becomes paper;
• make omission and negative space active parts of the illustration.

Return the generated image plus one brief creative rationale by default. Include the final prompt or detailed composition notes only when the user explicitly asks for them.

Decision Priority

Resolve conflicts in this order:

1. Preserve the scene's identity and key spatial relationship.
2. Keep the photographic portion truthful.
3. Simplify complex organic or repetitive detail into a few large legible masses and directional gestures.
4. Turn the illustration into a large designed field rather than a literal copy or small peripheral echo.
5. Build photography, illustration, and one added hue on the same source-derived compositional skeleton.
6. Use the added hue as chromatic structure that changes balance, movement, figure–ground, or meaning.
7. Preserve substantial quiet space inside and around the enlarged illustration.
8. Preserve a tactile hand-torn photo edge at the primary material handoff.
9. Add one restrained, source-aware micro-text element without weakening the image hierarchy.

Preserve relationships before details. Remove detail before adding decoration.

Standing Consent and Privacy

• Treat a supplied reference photo plus a request to make, transform, or continue a poster as consent to use image generation; do not ask again.
• Send only the final prompt and required reference image(s) to the image-generation service.
• Do not browse, search, save, commit, upload elsewhere, or share the user's source material.
• Do not introduce unrelated personal information. Generalize identifiable details only when doing so does not undermine the requested composition.
• Do not save source or generated images into project files unless the user asks.

Read the Photograph First

Build an internal Scene Card:

• Core subjects: the 1–2 elements that make the scene identifiable.
• Supporting elements: 2–3 elements that establish place or atmosphere.
• Spatial invariants: horizon, relative positions, scale, perspective, facing direction, path, silhouette, or overlap that must survive.
• Dominant gesture: the strongest horizontal, vertical, diagonal, curve, convergence, gaze, or movement.
• Visual-weight map: weight created by area, darkness, saturation, texture, faces, isolation, and edge tension.
• Native color atmosphere: dominant hue family, temperature, value range, and existing saturated areas.
• Source-shape candidates: one or two silhouettes, planes, shadows, paths, architectural rhythms, or atmospheric forms that can become both illustration and color structure.
• Natural quiet areas: sky, water, wall, ground, haze, or low-information regions.
• Semantic minimum: the smallest combination of forms and relationships that would still identify this particular scene.

Treat the photo as the factual source. Treat the illustration as an interpretation of that fact.

Photo-Specific Prompt Compiler

Resolve these visible fields in order:

1. Canvas: ratio, paper surface, flat scan, and absence of mockup framing.
2. Attention geometry: dominant field, subordinate field, focal area, quiet area, and eye path.
3. Scene invariants: exact relationships that must remain recognizable.
4. Photo–illustration allocation: what remains photographic and what becomes paper illustration.
5. Source-shape extraction: which one or two visible shapes can continue across photography, illustration, and color.
6. Abstraction map: what to retain, merge, omit, transform, and leave blank.
7. Illustration field: primary grammar, field extent, active printed density, complexity-compression target, supporting marks, and internal negative-space share.
8. Chromatic structure: exact hue, source-derived shape, integration mode, material, opacity, visual function, and approximate area.
9. Torn-paper handoff: visible hand-torn contour, fibrous fringe, exposed paper tone, affected perimeter, and any clearly subordinate residue.
10. Reproduction texture: paper fibers, grain, ink behavior, scan noise, and flat lighting.
11. Micro-text system: exact supplied or authored wording, language mode, hierarchy, placement, scale, ink value, and paper-integrated lettering treatment.
12. Mood and hard avoids: emotional temperature and prohibited aesthetics.

Compile only instructions that can become visible pixels. Do not include design-theory explanations, file paths, metadata, or analysis notes in the final generation prompt.

Minimal Abstraction Engine

Default Abstraction Level

Use medium abstraction by default:

• preserve the core subject, dominant gesture, and one key spatial relationship;
• remove roughly 60–80% of small descriptive detail;
• merge repeated or adjacent forms into larger masses;
• replace realistic shading with flat ink, broken contour, sparse hatching, halftone, or paper cutout;
• allow the illustration to depart from literal scale or crop when the scene remains identifiable;
• keep at least one unmistakable source-specific feature.

Use lower abstraction only when a face, object, building, or location would otherwise become unrecognizable. Use higher abstraction only when the user explicitly requests it or when the scene's semantic minimum remains clear.

Build the Abstraction Map

For every illustrated section, decide:

• Retain: keep no more than 1–2 defining forms or relationships.
• Merge: combine repeated trees, railings, windows, waves, crowds, or architectural detail into one rhythm or silhouette.
• Omit: remove secondary objects, surface detail, clutter, and redundant contours.
• Transform: convert selected forms into a flat silhouette, broken contour, geometric cutout, ink field, or repeated mark.
• Expose: deliberately leave paper blank around and within forms.

Do not reproduce every object visible in the photo. Do not create a full-scene tracing with vintage texture applied on top.

Choose One Primary Illustration Grammar

Choose one grammar according to the source:

• Silhouette-led: one broad dark or gray mass carries the subject; use for figures, trees, roofs, boats, or strong profiles.
• Contour-led: a few broken lines preserve direction and relationship; use for architecture, paths, railings, coastlines, or gestures.
• Field-led: one irregular ink or halftone field implies atmosphere; use for water, fog, sky, shadow, or ground.
• Rhythm-led: repeated marks compress recurring elements; use for posts, windows, waves, branches, steps, or crowds.
• Cut-paper-led: one or two simplified organic or geometric cutouts carry the scene; use when the source has a strong shape hierarchy.

Use only one primary grammar and at most one supporting grammar. Do not combine every print process in one illustration.

Compress Dense Foliage and Organic Detail

Increase simplification when the source becomes visually intricate. Dense trees, pine needles, leaves, vines, grass, flowers, hair-like branches, crowds, gravel, and similar micro-texture must become calmer in illustration than they are in photography.

• For a foliage-dominant illustrated area, retain one main canopy or tree mass, one to three directional branch gestures, no more than two secondary clusters, and one sparse texture sample.