OpenZhenZhen AI 提示词画廊 · MiniMax H3 · video
男友视角·走出空画框的人
Boyfriend's View: The One Stepping Out of the Empty Frame
男友视角·走出空画框的人，使用同名原始提示词配对并提供中英文版本
提示词 / Prompt
中文提示词
[参考生成] 15秒、16:9、真人写实、男友第一人称眼睛视角，一镜到底。<Subject 1>来自<Picture 1>，始终保留同一成年女性的脸、肤色、深色眼睛及深棕色长发；参考图片只用于身份与明确可见的衣着，不是本片开场帧。自然表演与真实物理，普通话对白按下述原句发声，不出现字幕或文字。 室内15秒，男友坐低椅，眼高约一米，从略低位置看两步外女方；她保持参考椭圆脸、白皙皮肤、深色眼睛及中分深棕长发，穿日常服装。宽大的独立落地肖像画框朝男友，外侧宽脚稳立；中间只有空洞、无画布玻璃，开口宽过她肩、高到成人能走过，底边只是贴地的细低条。框前男友一侧与右边均有通道；镜头同框看清她完整脸、开口和低条，不拍男友膝盖。首帧她双脚都在底条后，双手叠于腰前、肩平稳、视线略越过男友，故意像静止肖像，不先走动。她保持姿态问「要画，还是要人？」男友从坐姿答「要人。」听到选择后她才松开双手，直视他，自行一步跨过真实的低底条，连续穿过画框中间空洞，不穿过木边；两脚落在男友一侧，画框始终站稳。她再向画面右侧小移一步，让整个洞口真正变空，舒适距离处回头看他，作为真实的人而非静态肖像。结尾同一坐姿略仰视，她的脸与空框洞口同框；无切镜造成凭空消失。室内声、两句对白、跨低条与侧移脚步；无旁白、字幕或超自然变身声效。 仅指定人物和物件出现，动作因果与位置连续；男友的脸和身体不入画，只在需要时出现指定的手或鞋。环境声及对白与实际动作同步，无旁白、笑声罐头或额外音乐。
English prompt
subject_definitions: <Subject 1> is the adult woman in <Picture 1>. Preserve her recognizable oval face, fair skin, dark eyes and long center-parted dark-brown hair. Her everyday clothes, open portrait frame, stable outer feet and interior room are new episode elements. <Picture 1> supplies her identity, not an existing portrait or scene. summary: [reference generation] A continuous 15-second, 16:9 live-action story from her seated adult boyfriend's eyes. <Subject 1> holds an intentionally still portrait pose behind a large empty standing frame. Asked to choose between the picture and the person, he chooses her. She releases the pose, crosses through the actual open frame to his side, then shifts aside so the aperture is visibly empty while she looks back at him. retention_analysis: <Subject 1> (appears in [Shot 1]): fully_preserved - her facial identity and hair match the reference. The room, clothes and large freestanding empty frame are invented for this episode. detailed_description: mpov, first-person perspective pov video, low angle. One continuous 15-second 16:9 live-action view from my eyes while I remain seated on a low chair, eyes about one meter above the floor, looking slightly upward to <Subject 1> about two walking steps away. This is my eye position looking at her, never a separate camera in front of a couple. She begins entirely behind a broad freestanding portrait frame that faces me. The rigid frame stands on wide outer feet and stays put; its empty opening is wider than her shoulders and tall enough for an adult to step through. There is no picture, glass or canvas in the opening, and the bottom rail is a thin low strip against the floor. Clear landing space lies just in front of it on my side, with extra clear space to the right of its opening. My natural broad view sees her full face and the opening with its low rail; I do not need to look down at my own knees to establish the low viewpoint. [Shot 1] At the first frame <Subject 1>'s two feet are still behind the low rail. Her hands are folded before her waist, shoulders composed, gaze deliberately just past me: she is holding herself as a motionless portrait, not already walking out. Still in that pose, <Subject 1> (S1) asks, <d>[Chinese] 要画，还是要人？</d> I answer from my own seated position, me (S2), <d>[Chinese] 要人。</d> Only after hearing my choice does she unfold her hands, turn her eyes directly to mine and step forward herself. In one continuous visible passage she crosses the low bottom strip and comes through the empty center of the frame, never through its wooden side. Both feet land on my side while the frame remains standing behind her. She then takes one small step toward the right of my view, enough to clear the entire opening rather than still overlap it. She stops there at a comfortable distance, not pressing up to the lens, and looks back at me as a present person instead of holding the old portrait pose. End in the same gently upward seated eye line with her readable face beside a genuinely empty frame opening; do not cut from a person behind the frame to an unexplained already-empty frame. My viewpoint stays seated throughout; none of my face, torso or large knees enters the image, and there is no outside reverse shot. overall_soundscape: Quiet indoor ambience, her short Mandarin question, my nearby seated answer, one natural step over the low rail and a second small side step. No narration, captions, sound effect or music claiming a supernatural transformation. non_diegetic_music: N/A
主题与风格
探索更多图片和视频提示词 · 生成图片 · 生成视频 · 在 AI 画布中创作
相关作品
- 花朵绽放微距延时
生成一段20秒、16:9的花朵绽放微距延时视频。使用单一连续镜头，以平稳且极慢的推进展现花瓣逐渐舒展、雄蕊显露的过程。细致清晰地呈现植物纹理，在柔和自然光下保持真实色彩与宁静简洁的背景。全程让花朵保持清晰对焦；不加文字、不切镜、不添加其他物体。
- 暮色奢华套房漫游
生成一段20秒、16:9画幅的电影感豪华酒店顶层套房揭幕镜头。采用平滑的平视Steadicam运动，穿过关闭的双扇门持续推进，优雅连贯且不剪切。按层次展现套房：先展示铺有抛光大理石地面的入口大厅、洁白新鲜的兰花和闪耀的水晶吊灯；随后进入奶油色与金色装潢的客厅，家具整齐无瑕，落地窗外是暮色中闪烁的城市天际线，营造香港、纽约或迪拜的感觉，但不出现可见标识。镜头掠过冰桶中的香槟酒瓶，呈现真实的凝结水珠。穿过卧室门口，展示铺着洁白高支数床品并已掀开的特大床，枕头上放着一块巧克力。最后穿过敞开的推拉玻璃门来到阳台，展现下方灯火璀璨的城市。采用温暖精致的酒店灯光、逼真的材质、细腻反射、高端建筑摄影质感、稳
- 最后的园丁：地球重生
制作一部20秒、16:9画幅的写实科幻短片：一台约70厘米高、饱经风霜的小型原创园艺机器人，拥有圆润的旧金属机身、两条机械臂和一只圆形琥珀色眼睛，在被遗弃的地球上恢复生命。全片始终保持恰好一台机器人、一个金属浇水壶、一株幼苗和一株持续生长的植物，并保持机器人的比例及花朵位置一致。场景统一为龟裂干旱土地、毫无生机的土壤、尘雾和远处废墟。采用电影级构图、可触感的陈旧金属、真实植物生长、温暖金色时刻摄影和尘土氛围。时间线：0–3秒，低机位跟拍机器人独自行走，脚下干裂土地前方冒出一株微小绿苗；它停在幼苗旁，全程不剪辑。3–6秒，切镜后机器人跪下，用浇水壶把最后一滴水倒在土壤上，周围土地明显变湿。6–9
- 圣家堂环绕镜头
生成一段20秒、16:9画幅的电影感建筑视频。晴朗日光下，摄影机围绕圣家堂平滑连续地进行360度环绕运动，始终让教堂居中且清晰可辨；随着视角变化，展现其华丽塔楼、雕塑立面和精细建筑细节。保持地平线稳定、尺度一致、视差自然、石材纹理真实，并确保整个运动过程中的光照连贯。采用宽幅建立镜头，逐步展示建筑的多个侧面，运镜流畅且具有专业电影质感；不要剪辑，前景不要出现人物，不要文字或标志，也不要依赖音频呈现信息。
- 小径上的骏马由步行转为疾驰
生成一段20秒、16:9的电影感野生动物风格镜头：一匹马沿着自然土路小径前进。前10秒，马以稳定速度平静行走，清晰呈现完整身体，并展示真实的蹄部落点。约在中点时，马平顺加速，随后在剩余10秒沿同一条小径充满活力地疾驰。摄像机保持一致的侧后方跟拍角度，让马始终成为主体。使用自然日光、稳定曝光、真实的身体结构与运动，蹄部扬起轻微尘土，并呈现细致的户外环境。全程保持空间和时间连续性，不要剪辑、额外动物、人物、文字、标志或音频。
- 马匹小跑后饮水
MiniMax H3 视频生成规格。时长：严格 20 秒。画幅：16:9。制作一个写实的电影感场景：一间整洁、自然采光的马厩内，一匹健康的马围绕马厩持续小跑，清晰呈现有节奏的马蹄声视觉节拍和自然身体运动。画面中段，马逐渐减速，在饮水槽旁完全停下，低头开始饮水。保持连贯统一的镜头与轻柔跟拍，马匹解剖结构稳定，光线一致，环境细节自然。结尾让马平静地饮水；不要剪辑、文字、标志、品牌、源视频或音频输入。