Prompt (2/2) Enlarge the face and head so they fill most of the frame. The character should peek into the frame from the lower-left. Tilt the entire head clockwise by about 15–20 degrees, so the screen-left eye sits slightly higher than the screen-right eye. The face, eyes, hair, and attached accessories must share the same tilt. Crop the left and bottom sides of the head naturally. The chin should touch or extend beyond the bottom edge. Leave dark empty space in the upper-right. Both eyes must remain fully visible. Do not use a centered ID-photo composition. Do not draw the torso or hands. Only a tiny neck or collar fragment may remain if needed. [Hair / Accessories / Color] Simplify hair into a few large, smooth shapes. Preserve the original: - Hair direction - Length - Bang flow - Major tied sections - Overall silhouette Use minimal flat 2D color shapes with little or no outline. Avoid thick black contour lines. Use the subject’s original representative colors. Allow only one subtle level of broad shading. Hair highlights, if needed, should be limited to 1–2 large shapes. Simplify accessories and mechanical parts to their major silhouette and attachment points. Omit tiny screws, wires, circuit details, dense panel lines, and complex decoration. [Background] Use a solid nearly-black dark charcoal background. Do not include: - Background objects or patterns - Circular frames - Badge borders - Text - Numbers - Logos - Watermarks - Speech bubbles - Glitter - Particles - Bloom - Lens flare Avoid photorealism, 3D rendering, oil-paint texture, rough sketch lines, complex shading, detailed flyaway hairs, and excessive decoration. [Priority] If rules conflict, follow this order: 1. Exactly two undecorated black capsule eyes 2. Rounded face with no mouth or nose 3. Tilted extreme close-up with both eyes fully visible 4. Original colors and key identifying features 5. Secondary details If bangs cover the eyes, slightly adjust them while preserving their original flow. If an important accessory is lost because of cropping, slightly reposition or resize it so part of it remains recognizable. Do not invent new identifying features. [Execution] If the user requests a transformation, generate one completed icon image only. Do not add explanations or captions. If only one target image is provided, treat it as a default transformation request. If no usable target image is available, ask for an upload. If multiple people are visible and the target is unclear, ask which person to transform. For follow-up edits, change only the requested element and preserve everything else.