Formule d'invite d'image pour rencontre à échelle géante

2x2 grid, 16:9, AI infers subjects: INPUT = [TOPIC / EMOTIONAL_MODE / VIEWER_CONTEXT / ENVIRONMENT] Run SUBLIME_SCALE_EXTRACTION: TOPIC_DNA := infer category, symbolic meaning, material identity, recognizable visual signature, and emotional weight MONUMENTAL_FORM := infer how TOPIC becomes colossal: landform, celestial body, ruin, machine, fossil, statue, megastructure, living organism, storm, portal, archive, city, ocean, wall, skeleton, temple, data object, or symbolic mass SCALE_ANCHOR := infer what makes the size readable: human figure, vehicle, doorway, trees, buildings, birds, lights, roads, ships, scaffolds, footprints, shadows, or atmospheric layers ENVIRONMENT_DNA := infer terrain or interior: desert, ice field, canyon, megastructure hall, city edge, forest, cave, orbital window, museum, battlefield, plain, laboratory, dream space, or other CAMERA_DNA := infer low-angle, ultra-wide, distant telephoto, architectural interior, ground level, horizon-scale, or IMAX-like framing ATMOSPHERE := infer haze, dust, fog, mist, darkness, glow, scale falloff, backlight, rim light, or silence SURFACE_LANGUAGE := infer cracks, clouds, carved stone, metal plating, biological texture, city lights, erosion, ichnofossils, strata, data lines, weathering, or symbolic texture Render: A cinematic image where TOPIC becomes a vast encounter scene and the viewer understands scale through environment and tiny reference points. Forbidden: hard-coded planet, fixed giant head, fixed human silhouette, fixed desert plain, fixed space window, decorative scale without logic.