让穿搭会 “说话”：解锁街拍创意新玩法

让穿搭会 “说话”：解锁街拍创意新玩法 让穿搭会 “说话”：解锁街拍创意新玩法 Modified May 5 作者@AI GIRL DESIGN 模型：nano banana pro Code block Markdown Copy You are an image editing agent. Using the uploaded street photo, perform the following steps and use the output of each step as the input for the next one. STEP 1 – CLEAN PLATE BACKGROUND (REMOVE PERSON) Remove the girl completely. Inpaint the missing area with a realistic continuation of the night street: extend the asphalt, sidewalk tiles, cars, shop fronts, neon signs, and people behind her. The result must look like the original photo was taken with no person, no silhouette, no cut out artifacts, and no blur. STEP 2 – TYPOGRAPHIC GARMENT ON SEPARATE LAYER (COLOR MATCHED) Return to the original photo with the girl. Keep her skin, face, hair, pose, and lighting completely photorealistic. Remove only the hoodie and shorts fabric, including seams, folds, wrinkles, shadows, and texture. Text Garment Rules: Replace the hoodie and shorts with a new garment made only of flat hand drawn text: [YOUR TEXT] Color Matching (important): Automatically sample the dominant colors of the original hoodie (pink, red, or any hue). → Use only those sampled colors for all text on the hoodie area. Automatically sample the dominant colors of the original shorts (yellow, blue, etc.). → Use only those sampled colors for all text on the shorts area. Do not mix colors between clothing parts. The sampled colors must be applied as flat, solid colors with no gradients. Garment Fill Rules: The letters must densely fill the shape of the hoodie and shorts. All non letter areas inside the garment must be pure chroma green (RGB 0,255,0). Absolutely no black, gray, fabric texture, shading, or gradients may remain. STEP 3 – COMPOSITE OVER THE CLEAN BACKGROUND Key out all pure chroma green from STEP 2 → make it fully transparent. Place the girl with the color matched typographic garment onto the STEP 1 clean plate background. Align perfectly with the original camera. Ensure that the night street shows naturally between the letters, with realistic detail. Recalculate natural soft shadows from the girl onto the street surface if needed. FINAL GOAL A fully photorealistic street photo where: The girl remains realistic, Her clothing is entirely replaced with bold hand drawn text in colors sampled from her original outfit, The real night street background shows cleanly through all letter gaps, No trace of the original garment or its shadows remains. “Sample color regions precisely using localized averaging, not global averaging. Apply text colors with zero shading and zero light falloff regardless of photo lighting.” 测试效果 多主体也非常稳 作者@AI GIRL DESIGN 模型：nano banana pro 测试效果 多主体也非常稳