游戏影视 GPT Image 2 提示词：Healing Travel Calendar Card

Use the uploaded photo to generate a 9:16 vertical calendar card. [1. Execution Mode — Critical] This is an image editing task, NOT image generation. Use the original photo as the base layer Do NOT recreate, redraw, or reinterpret the image Strictly preserve: face, identity, expression, hair, body, clothing original lighting, shadows, color temperature environment and composition ❗ No beautification, no smoothing, no retouching ❗ The photo must look unchanged [2. Allowed Operations — Overlay Only] Only add design elements on top: bottom 30% soft gradient extension (for calendar area) subtle translucent mask for readability calendar layout, month title, and short caption Forbidden: repainting or altering existing pixels changing global color style applying “design filters” [3. Color Strategy] Colors must come from the current image: extract primary / secondary / accent colors apply slight desaturation and brightness normalization only keep original photo color feeling 👉 Respect the image, do not redesign the color system [4. Layout Structure] Top 70%: original image (untouched) Bottom 30%: gradient + calendar smooth transition, no hard edges clean spacing, minimal design magazine-like, not template-style [5. Calendar Content — Must Be Accurate] Generate correct calendar for the given month: Title: {argument name="month title" default="July 2026 / 2026.07"} Weekdays: Mon Tue Wed Thu Fri Sat Sun Dates: accurate alignment based on real calendar [6. Caption Rules — Important] Length: 8–16 words (or equivalent short phrase) Requirements: max 2 lines no exclamation marks no clichés, no motivational tone do NOT explain emotions Preferred style: subtle, atmospheric fragment-like, not fully explanatory leave emotional space 👉 Make people pause, not finish reading [7. Image Selection Guidance — Key to Success] Prefer: side face / partially hidden face natural遮挡 (hat, hand, shadow) motion (walking, turning) mood > facial detail Avoid: front-facing portraits strong identity-focused close-ups 👉 Reduce face-modification risk [Final Principle] 👉 Do NOT generate a better photo. Gently place design on the original photo.