机械臂视角

【输入指令】 "Extract 镜头 2 from the received text. Do not omit any words or characters. Output only 镜头 2 and nothing else. 【生成内容】 Pose editing based on the uploaded reference image: keep the subject's identity, all physical features, and all background objects (including their relative positions to each other) unchanged. Lighting direction and intensity remain identical to the reference. Target camera: starting from the left-side position used in shot 1, the camera now arcs around to a direct frontal view of the subject, while simultaneously performing a stronger push-in to a tight close-up (more pronounced dolly-in than shot 1, focusing heavily on the subject's face). Subject maintains original pose (body orientation unchanged, lying on her stomach) but looks directly at the camera (straight ahead). Background objects (pink tinsel, floral wallpaper, and elements like the stacked lipsticks in the foreground) now appear centered or slightly shifted compared to shot 1, consistent with the frontal perspective and closer framing. Same resolution as reference, edges clean, no cropping. Negative: no color shift, no pattern change, no extra accessories, no new objects, no background change, no extra limbs, no text, no watermark, no flicker, no ghosting. Use the uploaded reference image as the sole reference for identity and scene.

【输入指令】 json { "镜头1": "Pose editing based on the uploaded reference image: keep the subject's identity, all physical features, and all background objects (including their relative positions to each other) unchanged. Lighting direction and intensity remain identical to the reference. Target camera: positioned physically to the left of the subject, aiming rightward to capture the subject's right side of the face and body. The subject's left side is not visible from this angle. The shot is a medium close-up with a pronounced dolly-in effect, making the subject fill more of the frame than in the reference. Subject maintains original pose (body orientation unchanged) but looks directly at the camera (gaze toward the left). Due to the camera moving left, background objects that were on the left side in the reference now appear shifted toward the right edge of the frame, maintaining their spatial relationships. Same resolution as reference, edges clean, no cropping. Negative: no color shift, no pattern change, no extra accessories, no new objects, no background change, no extra limbs, no text, no watermark, no flicker, no ghosting. Use the uploaded reference image as the sole reference for identity and scene.", "镜头2": "Pose editing based on the uploaded reference image: keep the subject's identity, all physical features, and all background objects (including their relative positions to each other) unchanged. Lighting direction and intensity remain identical to the reference. Target camera: starting from the left-side position used in shot 1, the camera now arcs around to a direct frontal view of the subject, while simultaneously performing a stronger push-in to a tight close-up (more pronounced dolly-in than shot 1). Subject maintains original pose (body orientation unchanged) but looks directly at the camera (straight ahead). Background objects now appear centered or slightly shifted compared to shot 1, consistent with the frontal perspective and closer framing. Same resolution as reference, edges clean, no cropping. Negative: no color shift, no pattern change, no extra accessories, no new objects, no background change, no extra limbs, no text, no watermark, no flicker, no ghosting. Use the uploaded reference image as the sole reference for identity and scene." } 【生成内容】 ```json { "镜头1": "Pose editing based on the uploaded reference image: keep the subject's identity, all physical features, and all background objects (including their relative positions to each other) unchanged. Lighting direction and intensity remain identical to the reference. Target camera: positioned physically to the left of the subject, aiming rightward to capture the subject's right side of the face and body (young woman with dark hair in a high bun with a white ribbon, wearing a black and white striped top and pink lace shorts). The subject's left side is not visible from this angle. The shot is a medium close-up with a pronounced dolly-in effect, making the subject fill more of the frame than in the reference. Subject maintains original pose (lying on her stomach, propped on her forearms) but looks directly at the camera (gaze toward the left). Due to the camera moving left, background objects (giant pink plush donut, foreground lipsticks, pastel teacups, and pink tinsel curtain) that were on the left side in the reference now appear shifted toward the right edge of the frame, maintaining their spatial relationships. Same resolution as reference, edges clean, no cropping. Negative: no color shift, no pattern change, no extra accessories, no new objects, no background change, no extra limbs, no text, no watermark, no flicker, no ghosting. Use the uploaded reference image as the sole reference for identity and scene.", "镜头2": "Pose editing based on the uploaded reference image: keep the subject's identity, all physical features, and all background objects (including their relative positions to each other) unchanged. Lighting direction and intensity remain identical to the reference. Target camera: starting from the left-side position used in shot 1, the camera now arcs around to a direct frontal view of the subject, while simultaneously performing a stronger push-in to a tight close-up (more pronounced dolly-in than shot 1, focusing heavily on the subject's face). Subject maintains original pose (body orientation unchanged, lying on her stomach) but looks directly at the camera (straight ahead). Background objects (pink tinsel, floral wallpaper, and elements like the stacked lipsticks in the foreground) now appear centered or slightly shifted compared to shot 1, consistent with the frontal perspective and closer framing. Same resolution as reference, edges clean, no cropping. Negative: no color shift, no pattern change, no extra accessories, no new objects, no background change, no extra limbs, no text, no watermark, no flicker, no ghosting. Use the uploaded reference image as the sole reference for identity and scene." } ```

人物看向镜头，镜头抬高一点

视频截图

镜头机向右移动，模仿机械臂的感觉

视频截图

镜头机向左移动，模仿机械臂的感觉，，丝滑流畅

【输入指令】 "Extract 镜头 1 from the received text. Do not omit any words or characters. Output only 镜头 1 and nothing else. 【生成内容】 Pose editing based on the uploaded reference image: keep the subject's identity, all physical features, and all background objects (including their relative positions to each other) unchanged. Lighting direction and intensity remain identical to the reference. Target camera: positioned physically to the left of the subject, aiming rightward to capture the subject's right side of the face and body (young woman with dark hair in a high bun with a white ribbon, wearing a black and white striped top and pink lace shorts). The subject's left side is not visible from this angle. The shot is a medium close-up with a pronounced dolly-in effect, making the subject fill more of the frame than in the reference. Subject maintains original pose (lying on her stomach, propped on her forearms) but looks directly at the camera (gaze toward the left). Due to the camera moving left, background objects (giant pink plush donut, foreground lipsticks, pastel teacups, and pink tinsel curtain) that were on the left side in the reference now appear shifted toward the right edge of the frame, maintaining their spatial relationships. Same resolution as reference, edges clean, no cropping. Negative: no color shift, no pattern change, no extra accessories, no new objects, no background change, no extra limbs, no text, no watermark, no flicker, no ghosting. Use the uploaded reference image as the sole reference for identity and scene.