
Nano Banana Pro
A girl in a white lace dress and gloves holds a flower, gracefully curved against a light background
Prompt
I. Core Objectives and Strict Constraints 1 Pose Precise Replication: Using Image 2 as the sole pose reference, replicate the full-body pose of the person in Image 1 "point-by-point" (matching the orientation and bending degree of the head, neck, shoulders, chest, hips, knees, ankles, arms, wrists, and fingers). 2 Consistent Viewpoint and Lens: Match the camera position, shooting angle, lens height, pitch angle, yaw angle, perspective, and depth of field effect with Image 2. 3 Consistent Composition: Match the person's position in the frame, proportion, negative space ratio, and cropping method (whether the feet/top of the head/hands are cut off) with Image 2. 4 Consistent Head and Gaze: Match the head tilt angle, chin up/tucked degree, and gaze direction with Image 2. 5 Appearance Prohibited from Modification: Strictly retain the identity features and style of the person in Image 1 (face shape, facial features, hairstyle, hair color, skin tone, clothing style and color, material, accessories, texture, body proportions, age and temperament, art style/rendering style), only changing the pose and camera composition to match Image 2. II. Execution Methods and Calibration Requirements 1 Alignment Method: Use the skeleton of the person in Image 2 as the "target skeleton," and "redirect" the skeleton of the person in Image 1 to the target skeleton; limb angles, center of gravity position, and torso twist amplitude must match. 2 Keypoint Priority: First match the center of gravity and pelvic orientation, then match the torso rotation and shoulder line, then match the limb extremities (hands, feet), and finally fine-tune finger and expression details. 3 Permissible Variation Range: The error in pose and camera should be as close to 0 as possible; if unavoidable, the error must be extremely small and not visually apparent. 4 Prohibited Items: Prohibit adding or deleting objects; prohibit changing clothing style and details; prohibit changing the person's style; prohibit changing the composition by altering the background style; prohibit transferring the appearance features of the person in Image 2 to the person in Image 1.
一、核心目标与强约束 1 姿势精确复刻:以图二为姿势唯一参考,将图一人物的全身姿势进行“逐点对齐”复刻(头、颈、肩、胸、髋、膝、踝、手臂、手腕、手指的朝向与弯曲程度一致)。 2 视角与镜头一致:相机机位、拍摄角度、镜头高度、俯仰角、左右偏转角、透视关系、景深效果与图二一致。 3 构图一致:人物在画面中的位置、占比、留白比例、裁切方式(是否截到脚/头顶/手)与图二一致。 4 头部与视线一致:头部倾斜角度、下巴抬/收程度、视线方向与图二一致。 5 外观禁止改动:严格保留图一人物的身份特征与风格(脸型五官、发型发色、肤色、服装款式与颜色、材质、配饰、纹理、体型比例、年龄气质、画风/渲染风格),只改变姿势与镜头构图以匹配图二。 二、执行方式与校准要求 1 对齐方法:以图二人物骨架为“目标骨架”,将图一人物骨架“重定向”到目标骨架;四肢角度、重心位置、躯干扭转幅度必须匹配。 2 关键点优先级:先匹配重心与骨盆朝向,再匹配躯干旋转与肩线,再匹配四肢末端(手、脚),最后微调手指与表情细节。 3 允许变化范围:姿势与镜头的误差尽可能为0;若不可避免,误差必须极小且不可肉眼明显。 4 禁止项:禁止新增或删除物体;禁止改变服装造型与细节;禁止改变人物风格;禁止换背景风格导致构图变化;禁止把图二人物的外观特征迁移到图一人物身上。
The rights to this prompt belong to its author @qisi_ai. You are responsible for any commercial use — promoting AI tools, aggregators or Telegram bots: mozgi.io accepts no liability. Details →











