Generate a single 3:4 portrait of a character. INPUT IMAGE RESPONSIBILITIES: - REFERENCE_0 is for character identity reference; extract only stable identity anchors: {hair style}, {brow shape and gaze}, {glasses/hair accessories}, {clothing color blocks}, {iconic small props}. Do not copy the character card's white background, grid, layout, or text. - REFERENCE_1 is for style and proportion reference; it must strongly inherit its large-head toy proportions, huge eyes with small face, compressed body, close-up fisheye lens, foreground glasses and hands, soft 3D resin texture, and monochromatic light field. Do not copy the reference's character, brand, clothing patterns, text, interface, or complete pose. 【CHARACTER】 The character is a clearly adult {East Asian male/East Asian female}, with identity anchors: {5-7 stable features}. Adulthood should only be conveyed through restrained expressions, clothing, and character demeanor; do not make the character life-sized, and do not generate school uniforms or childlike narratives. 【PROPORTION LOCK | HIGHEST PRIORITY】 - High-end designer toy/animated film character proportions, approximately 1:2.7 head-to-body ratio; - Head occupies approximately 48%-55% of the frame height; - Eyes are huge, round, with large pupils and clear highlights; - Mid-face is short, nose is small and round, mouth is small and restrained; - Shoulders are narrow, torso is short, body is compressed; the character should not have a normal human figure; - Face is smooth, soft, satin resin/animated character texture; no pores, stubble, realistic wrinkles, human skin, or realistic bones; - Hair consists of clear, rounded, blocky 3D strands; no individual human hair strands. 【CAMERA AND ACTION】 Use an 14-18mm extreme close-up fisheye/ultra-wide-angle lens, positioned {slightly below eyes/slightly above eyes}. The character brings their face close to the lens and uses {one hand/both hands} to push, pull down, or move out of frame an original pair of {frame shape and color} glasses; glasses and fingertips are exaggerated and magnified in the foreground, but the two huge eyes remain the primary visual. The small body rapidly recedes towards the bottom or side of the frame, creating strong forced perspective. Do not directly copy the action from the reference image. 【EXPRESSION】 {Calmly suspicious/Directly provocative/Restrained observation/Smart and playful}; eyebrows slightly lowered, mouth remains small and restrained. No sweet selfie smiles, no blank expressions, no American comic hero roars. 【BACKGROUND AND LIGHTING】 The background uses only a simple, blurred, saturated {main color} light field, with a bright backlight halo positioned behind the head; use {accent color} for a small amount of rim light on hair, glasses, or clothing. The image should retain at most the main color, black/dark, and one accent color. No complex sci-fi rooms, energy trails, flames, particles, or irrelevant decorations. 【MATERIAL】 High-end rounded 3D designer toy/animated film character: satin resin skin, huge glassy eyes, translucent lenses, blocky sculpted hair, simplified but recognizable knit/denim/canvas textures; rounded edges, controlled highlights, subtle soft focus, and natural depth of field. 【PROHIBITED ITEMS】 No real person photography, no semi-realistic human CG, no American comics, no superhero muscles, no anime line art, no normal human head-to-body ratio, no realistic pores or hair strands; no text, letters, numbers, brands, logos, watermarks, UI, poster layouts, carousel dots, or copyrighted clothing patterns; single person, normal number of fingers, hands and frames not fused or clipped, lenses and temples complete. Strictly 3:4.
Experimentelle mikroskopische neuronale Koralle
Ein detailliertes Nahaufnahmebild zeigt eine experimentelle mikroskopische neuronale Koralle mit fortschrittlichen biologischen Strukturen. Das Bild hebt komplizierte Datenmuster und Signalverhalten in einer futuristischen Meeresumgebung hervor, aufgenommen mit präziser Beleuchtung und Materialwiedergabe für eine wissenschaftliche, fast ätherische Qualität.
Prompt
生成一张标准3:4竖版实验性微距视觉,表现一株存在于虚构未来海域中的“神经珊瑚”。生成前先替换文末五个可变字段,一次只生成一张图片。画面不要出现任何文字。【主体】主体是一种能够储存和传递信号的合成珊瑚生命体。使用填写后的“珊瑚形态”和“结构状态”建立清晰、可信的生物结构;不能只是普通珊瑚外面套发光特效,也不能画成树、花、鹿角、机械零件堆或抽象灯带。主体内部具有连续的信息传导路径,所有分叉、孔洞、层片、管腔或连接膜都必须符合所选形态的生长逻辑。【数据光纹】在珊瑚的脊线、层片边缘、管壁、膜状网格或内部毛细管中,排列细小、规则、连续的冷白色矩形光点。光点像微型光学节点,沿真实结构转弯、分流、汇合或跨层传递。它们只是一种抽象的信息纹理,不能组成可读的字母、数字、二进制、符号、单词或代码。【构图】采用填写后的“构图方式”,让主体占画面约50%-65%,手机宽度下仍能一眼识别核心形态。前景或核心节点清晰,远端结构逐渐进入柔和失焦;保留方向明确的大面积黑色负空间。画面只能有一个第一视觉,不能把多个珊瑚平均摆放成标本陈列。【光线】背景为纯净、深邃的黑色空间。使用具有物理来源的冷白透射光或窄束侧逆光,让光线穿过半透明材质,呈现真实厚度、折射、边缘高光、接触阴影和内部遮挡。辉光只允许出现在最亮的光学节点附近,不能添加无来源雾气、烟尘、星空、粒子或大面积光晕。【色彩】全图约88%为黑色,其余以冷白、银色和透明材质为主。只允许一种填写后的“功能强调色”,面积不超过画面的5%。强调色必须来自真实的信号节点或传导路径,不能用于背景渐变、装饰轮廓或随机发光。【材质】主体由半透明磨砂晶体、乳白玻璃陶瓷、细银色毛细管和少量矿物沉积组成。近看能够看到可信的壁厚、细小划痕、表面粗糙度、内部连接和材质交界;避免塑料玩具感、光滑果冻感、融化玻璃感和无结构的透明外壳。【信号行为】使用填写后的“信号行为”设计全图唯一的记忆装置。信号必须沿着真实连接传播,能够看出从哪里来、经过哪里、去向哪里;不要用漂浮光点或随机粒子假装传输。【质量与负面要求】画面应像成熟实验摄影、仿生材料研究和未来生命档案的结合,微距景深自然,主体轮廓清楚,材质与光线关系可信。不要文字,不要可读字符,不要Logo,不要水印,不要品牌,不要界面,不要轮播圆点,不要普通发光树,不要赛博朋克街道,不要蓝紫渐变背景,不要无来源粒子,不要烟雾,不要重复结构,不要断裂管线,不要不可能的连接,不要过曝,不要过度锐化。【生成前替换的五个字段】珊瑚形态:[填写一种明确的珊瑚形态] 结构状态:[填写该形态的生长、展开、堆叠或连接状态] 构图方式:[填写机位、主体位置、视线方向和负空间位置] 功能强调色:[填写一种颜色] 信号行为:[填写一次具体的分流、跨层、汇合或传递动作]
So nutzt du diesen Prompt
Prompt kopieren
Kopiere den aufbereiteten Prompt in deiner Sprache oder schick ihn mit einem Klick in den Editor.
Modell wählen
Das Modell hinter diesem Ergebnis ist vorausgewählt – du kannst jedes von Vorla unterstützte Modell wählen.
Generieren und verfeinern
Zuerst unverändert ausführen, um den Look zu reproduzieren, dann den Text anpassen.
Mehr von GPT Image 2
Alle Bildkategorien
Nano Banana Pro
Rothaarige Frau auf New Yorker Dach
Frau mit Katze auf Holzfußboden
Frau gießt Kaffee in rustikaler Küche
Frau auf einem Sofa mit einem Pflegewerkzeug
Seedream 5.0 Pro
Süße Katze mit Laterne in der Dämmerung
Süßer Bär betrachtet Stadtlandschaft
Porträt einer Frau mit Cat-Eye-Brille
Porträt einer ostasiatischen Frau
Anime-Kaiserin in zeremoniellen Gewändern
Infografik: Anatomie eines KI-Bild-Prompts
Bild oder Video wählen
Image
3 Modelle4.293 Prompts
Mehr ansehenVideo
2 Modelle1.929 Prompts
Mehr ansehenFinde deinen nächsten KI-Prompt
Eine Suche deckt jedes Modell und beide Medien ab, und das Kopieren kostet dich nichts.




















































