OpenZhenZhen AI 提示词画廊 · gpt-image-2.5 · image

无边框七层微缩叙事世界

Borderless Seven-Layer Miniature Narrative World

以高角度等距视角呈现一条自下而上蜿蜒的七层微缩叙事路径；前景地形延伸至画幅底边，巨型3D作品标题直接立于地表，同一人物贯穿每个故事节点。

无边框七层微缩叙事世界 — AI image creation
无边框七层微缩叙事世界

提示词 / Prompt

中文提示词

使用已配置的身份参考ref-10a-v1，让同一位明确成年的角色在整个场景中反复出现且身份一致。以高角度等距视角创作作品“守灯人”的高密度微缩世界，地形表面连续铺满整张横向16:9画幅并延伸至四边。画面底边必须呈现继续向画外延伸的地表顶面；绝不出现底板、盒子、木框、玻璃罩、垂直地层剖面或切面。沿一条清晰易读的上升之字形路径，按顺序构建恰好七个不同叙事层次，并依照以下所给故事节点顺序呈现：1）在海边村庄发现发光地图；2）穿过古老石桥；3）跟随萤火虫进入森林；4）在山间天文台避风；5）在雪岭点亮灯笼；6）引导船只驶过雾港；7）眺望远方城市日出。在最靠近观者的底部前景，将巨大的电影感3D立体字“守灯人”直接竖立在地形表面，文字在同一地表投下可信阴影。让守灯人作为重复出现的微缩角色贯穿每一层，并逐层执行对应故事节点中的动作；每次出现都保持与ref-10a-v1一致的身份。地表材质与场景随所给节点协调变化，同时让整个微缩世界保持视觉连贯。使用紧凑但清晰可辨的微缩环境、自然等距纵深、移轴微距摄影观感、带触感的黏土动画式材质细节和电影感体积光。整体如同更大现实世界中无限延伸的一片切面，地形必须出血至画幅四边。标题须拼写准确、清晰可读，并自然融入地表。不要边框、容器、底板、剖面、切面、额外文字、所给七个节点以外的场景、身份不一致的重复角色、无关人物、标志或水印。

English prompt

Use configured identity reference ref-10a-v1 for the same clearly adult character, repeated consistently throughout the scene. Create a hyper-dense, high-angle isometric miniature world representing THE LANTERN KEEPER, with one continuous terrain surface that fills the entire horizontal 16:9 frame edge to edge. The bottom edge must show the top surface of the terrain continuing off-screen—never a baseplate, box, wooden frame, glass case, vertical ground cross-section or cutaway. Build a readable rising zigzag path through exactly seven distinct narrative layers, using these supplied story beats in order: 1) finds a glowing map in a seaside village; 2) crosses an old stone bridge; 3) follows fireflies through a forest; 4) shelters in a mountain observatory; 5) lights a lantern on a snowy ridge; 6) guides boats through a foggy harbor; 7) watches sunrise over the distant city. At the immediate bottom foreground, plant massive cinematic 3D typography spelling the exact work title “THE LANTERN KEEPER” directly on the terrain; cast convincing shadows from the letters onto that same surface. Show the Lantern Keeper as the recurring miniature figure in every layer, performing the matching action from the corresponding supplied beat; preserve the same ref-10a-v1 identity at each appearance. Make the surface material and setting change coherently with the supplied beats while keeping the whole world visually connected. Use tightly packed but legible miniature environments, natural isometric depth, tilt-shift macro-photography cues, tactile claymation-like surface detail and cinematic volumetric light. The world should feel like an infinite slice of a larger reality, with terrain visibly bleeding to every image edge. Keep the title readable, correctly spelled and integrated into the terrain. No borders, container, base, cross-section, cutaway, extra text, scene outside the seven supplied beats, duplicate character identities, unrelated figures, logos or watermark.

查看 API 模型

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