پرامپت · ۱
2x2 grid, 16:9, 4 AI inferred dystopian subjects,topics: // Canvas grid 600x800 (vertical panel). Each cell encodes ink level + style ownership.
// 1. Photo ink-ification:
// - For each cell, compute luminance from [Location Plate] photo.
// - L > 200 -> paper-white (0). L < 80 -> solid black (1). Else -> mid-tone (2).
// - Mid-tone cells are patterned with a halftone dot array at declared LPI.
// - Edge detection: cells where luminance gradient exceeds threshold become mechanical line (3) — uniform thickness, no taper.
// 2. Character rendering:
// - Using [Char Seed], place a small figure (under 25% cell area) in the scene.
// - The figure cells are drawn in the dominant style: line owned by dominant, flats by dominant, light by dominant.
// - Subordinate style may appear in background elements (props, textures) but never on character face cells (defined by a mask). If dissolved, subordinate elements are redrawn with dominant line. If declared, boundary cells between styles are explicitly a different color. If oscillating, subordinate only in depth planes behind the character (assigned by depth map).
// - Silhouette anchor: hair shape + accessory preserved in pure black at any scale.
// - Color key: accent color appears only in eye cells and one prop cell cluster.
// 3. Interaction:
// - For tourist: character's contact cells have a shadow cluster offset according to sun vector (e.g., shadow cells 5 pixels down-right).
// - For fugitive: mid-tone clutter cells from the photo are placed in front of the character partially (overlap).
// - For haunting: no shadow, no interaction.
// 4. Grain-and-artifact: a layer of noise cells (JPEG artifacts, dust) is added only to photo-derived cells, not character.
// 5. Lens distortion: at edges of the canvas, character cells are slightly displaced to simulate barrel distortion matching the photo.
// Render final cell grid as high-res ink-ified image.
