O que é o GPT-Image-2?

O GPT-Image-2 é o modelo de geração de imagens da OpenAI e no Labnana suporta 1K, 2K e 4K. Ele segue bem instruções estruturadas e vai bem em ilustração limpa.

Como uso estes prompts do GPT-Image-2?

Escolha «Gerar» em qualquer card e o modelo e o prompt entram no gerador. O texto do prompt também pode ser copiado sozinho.

O GPT-Image-2 é gratuito no Labnana?

Dá para usar com conta gratuita. As gerações gratuitas do GPT-Image-2 vêm de campanhas e o menu de modelos mostra quantas restam; depois disso a cobrança passa para créditos universais, com 5 a cada check-in diário. 4K exige plano pago.

Como estruturar um prompt do GPT-Image-2?

Separe sujeito, composição, estilo e luz. Como o modelo segue de perto instruções estruturadas, deixar cada exigência explícita aproxima o resultado da intenção; comparar o texto de um card com a imagem ajuda a calibrar.

GPT-Image-2 ou Nano Banana?

Use o GPT-Image-2 quando o resultado precisa bater com uma especificação escrita ou o estilo é de ilustração. Use o Nano Banana para tipografia dentro da imagem, consistência de personagens e detalhe fotorrealista. O jeito mais direto é rodar o mesmo prompt nos dois.

O GPT-Image-2 consegue escrever texto dentro da imagem?

Consegue, desde que o texto seja curto e esteja entre aspas no prompt. Para tipografia densa ou muito estilizada, o Nano Banana Pro se sai melhor.

Quais resoluções o GPT-Image-2 suporta?

1K, 2K e 4K. 1K e 2K estão liberados no plano gratuito; 4K exige plano pago.

Estes prompts são gratuitos?

São: foram publicados pela comunidade e copiá-los ou adaptá-los não custa nada.

Posso usar os resultados comercialmente?

Depende das políticas de uso do fornecedor do modelo. O Labnana não coloca marca d’água; confira essas políticas antes de usar as imagens em trabalho remunerado.

Labnana Entrar

1 imageGPT-Image-2

GPT-Image-2 · OpenAI

Prompts do GPT-Image-2

O modelo de imagem da OpenAI no Labnana, com o prompt que produziu cada resultado.

See prompt examples

Obediência fiel às instruções
Ilustração limpa
Condução por imagens de referência
Saída em alta resolução

Extract character, anime CG, with a slight hand-drawn style.
Capture the moment the character look — Real GPT-Image-2 generations — open one for the full prompt

竖版插画，复古手绘板风格，非厚涂，高对比度。整幅画摒弃平滑渐变，以老旧羊皮纸般的斑驳肌理、丝网印刷的粗糙网点与干涩的刮擦排线构成全部影调。画面色彩极度克制，仅使用深邃的暗黑与大面积高饱和的金黄色，两种 — Real GPT-Image-2 generations — open one for the full prompt

O GPT-Image-2 é o modelo de geração de imagens da OpenAI e no Labnana suporta 1K, 2K e 4K. Seu diferencial é a obediência a instruções estruturadas: quanto mais claros estiverem sujeito, composição, estilo e luz, mais o resultado se aproxima do pedido.

Todos os prompts abaixo vêm de gerações publicadas e correspondem um a um à sua imagem. Ao escolher «Gerar», o prompt, o modelo e os parâmetros originais entram no gerador no topo da página.

Using GPT-Image-2 on Labnana

Escolha o canal de geração

O GPT-Image-2 fica no grupo OpenAI do menu de modelos e oferece um canal normal e um canal Stable. Se a consistência entre gerações importa mais, escolha o Stable.

Escreva as exigências de forma estruturada

Este modelo aproveita melhor entradas estruturadas. Separe sujeito, composição, estilo e luz para que cada exigência chegue inteira ao resultado.

Resolução e proporção

1K, 2K e 4K em 10 proporções, de 1:1 a 21:9. Contas gratuitas usam 1K e 2K; 4K exige plano pago.

Imagens de referência

Anexe até 5 referências por geração para levar adiante um estilo, um sujeito ou um layout sem precisar descrevê-los de novo a cada prompt.

Cota gratuita no GPT-Image-2

As gerações gratuitas do GPT-Image-2 são distribuídas por campanhas da plataforma, e o menu de modelos mostra em tempo real quantas restam na sua conta.

Cota de campanha: Distribuída por campanhas; o saldo aparece no gerador
Check-in diário: 5 créditos universais, válidos em qualquer modelo
Resoluções gratuitas: 1K e 2K: primeiro consomem as gerações gratuitas, depois os créditos universais
Limite de resolução: 4K exige plano pago e a cota gratuita nunca vale para 4K
Marca d’água: Nenhuma, no plano gratuito ou pago

Se você precisa de uma cota gratuita fixa que não dependa de campanhas, use o Nano Banana Pro: 2 gerações no cadastro e mais 1 a cada check-in diário.

Ver planos e créditos

Prompts populares do GPT-Image-2

Ordenados pela popularidade atual das gerações publicadas com o GPT-Image-2.

GPT-Image-2

Extract character, anime CG, with a slight hand-drawn style. Capture the moment the character looks at the camera from the side. Character height 148 cm The pendant accessory on the character's chest is three-dimensional and spherical. Vividly showcase the character's temperament. No text appears in the image.

Detalhes

GPT-Image-2

竖版插画，复古手绘板风格，非厚涂，高对比度。整幅画摒弃平滑渐变，以老旧羊皮纸般的斑驳肌理、丝网印刷的粗糙网点与干涩的刮擦排线构成全部影调。画面色彩极度克制，仅使用深邃的暗黑与大面积高饱和的金黄色，两种色彩的对撞构成最极致的视觉剥离。绝对居中的神龛式构图图一的骑士少女处于画面正中央，身体轴线与中轴线完全重合。她正身面对镜头，双臂微微张开，形成一个对称的、略带仪式感的开放姿态。她的背后升起一面巨大的、类似太阳的半圆形光环，光环以粗犷的手绘金线勾勒外轮廓，内部填充高饱和度的金黄色。光环表面布满三角形几何图纹，三角形以断续的深黄与留白线条交替排列，从光环中心向外辐射，如同某种被遗忘的宗教图腾。这面光环将她从头部到肩部完整框住，形成一个类似祭坛画或神龛的视觉结构，将观者的视线强制锁定在她面带微笑的脸庞上。光环的金黄色并非干净的平涂，而是布满了模拟老旧羊皮纸的破损斑驳、涂抹污渍与不均匀的油墨渗痕，边缘有灼烧般的深褐色焦痕网点，仿佛这面光环是一件被时间与火焰舔舐过的古老圣物。背景中的哥特式尖塔群在光环之外的纯黑背景中，隐隐浮现哥特式教堂的尖塔群。尖塔以断续的深灰与暗金轮廓线勾勒，线条向上收拢、锐利而笔直，直指画面顶端。塔身有极简的竖长窗格与飞扶垛轮廓，以稀疏的网点填充暗部。尖塔群排列在画面中轴两侧，越靠近边缘越高耸，形成一道垂直向上的视觉引导，强化了宗教建筑的庄严、压迫与向上的空间纵深感。塔尖在画面顶端几乎刺破画框，将整幅画面的情绪推向一种狂热、压抑的献祭氛围。左右对称的文字排版画面左侧竖排一行醒目的拉丁文主标题，字母以粗糙的手写大写体呈现，笔触模仿干涸的马克笔或旧式印刷字模，边缘有飞白与轻微渗墨。文字为“FINIS MYTHOLOGIAE”，字体粗黑有力，每个字母的间距均等，纵向排列，压在左侧从光环边缘到画面底边的空间内。标题上方绘有一个极简的耶稣鱼符号——两道交叉的弧线以断续的粗线勾勒，内部留空——以及一个圆形的太阳符号，太阳以金黄色平涂，边缘有放射状短排线。画面右侧对应位置竖排一行副标题，字体略小于左侧，同样是手写大写体，内容为“ET LUX PERPETUA”或类似短句，副标题下方绘有一个倒置的十字架，以粗黑断续的线条勾勒，十字架横臂两端微微上翘，在黑暗童话语境中带有异端或反叛的符号暗示。左右文字与符号的对称排布压住了画面的两侧，形成一本摊开的古老经卷般的庄重视觉结构。金白发骑士少女的异色凝视与危险微笑她的身形以近乎纯黑的剪影处理，仅保留极少量的灰色细节来区分肢体与铠甲的层次。一头金白色的长发从肩头垂落，发色以纸白留白为主，局部铺以极淡的黄色网点，发丝边缘以断续的细线勾勒，在纯黑的躯体剪影上划出几道锐利的光痕。长发在光环的金黄背景中形成微弱的银白对比，如同被圣光漂白的细丝。她的面部轮廓以笃定的粗黑线勾勒，皮肤以纸白平涂，不施加任何柔和阴影，仅在颧骨下方与下颌缘用稀疏的深灰网点压出微弱的立体感。她的眼睛是红蓝异色瞳——左眼虹膜以高饱和度的深红色平涂，右眼虹膜以深蓝色平涂，两颗瞳孔均以针尖大小的墨点表现，正直直地望向镜头。双眼微微半眯，上眼睑以两条平缓的弧线压下，遮住虹膜的上半部分，睫毛以数根笃定的细线向外翘起。这个半眯的眼神不是困倦，而是一种慵懒的、审视猎物般的专注。她的嘴唇微启，嘴角上扬，露出一个浅浅的微笑。嘴唇以粗黑的断续线勾勒，上唇薄而平直，下唇微微饱满，嘴角的弧度极小却异常清晰，牙齿以几道极细的留白短线暗示。这个微笑配合半眯的异色瞳，传递出一种安静的、自信到近乎轻蔑的危险感——她不是在威胁，她只是在欣赏某个即将发生的事情，或者某个人即将到来的反应。她的双臂微微张开，手掌向下，手指自然分开。左手倒握着一把佩剑，剑柄朝下，剑尖朝上，剑身以纯黑的色块表现，仅在剑刃一侧用一道极细的留白线勾出锋口。右手则空手张开，五指伸展，手指以粗黑轮廓线勾勒，指甲以极细的留白弧线轻描。她的双臂展开的姿态带着某种仪式的意味——不是投降，不是拥抱，而是一种展示，一种宣告，仿佛她正在说：“看，这就是结局。” 黄黑剪影的极致对比与服饰暗纹她的身体与服饰完全以纯黑色块剪影表现，没有内部的光影渲染。铠甲的轮廓以笃定的粗线勾勒，胸甲有微微隆起的弧度，肩甲方正而硬挺，腰际有束带的轮廓线，裙甲以层叠的短弧线分层。黑色剪影内部仅在最必要的转折处——肩甲的边缘、胸甲的中心线、束带的扣环——以极细的留白线或浅灰网点轻描，形成微弱的灰部细节，如同版画中刻刀划过黑色油墨留下的刻痕。这种极度扁平的纯黑处理，与身后高饱和、斑驳的金黄色光环形成了所有视觉关系中最极致的剥离——她不是一个“有血有肉”的人，她是一个符号，一个被烙印在羊皮纸经卷上的审判者剪影。斑驳肌理与版画质感整幅画面的金黄色区域布满不均匀的油墨渗痕、细小的黑色杂点、模拟丝网印刷套色不准的微偏移以及故意保留的涂抹刮擦纹理。光环的三角图纹中有墨水干涸后的龟裂纹以极细的排线表现。黑色区域同样不纯净——纯黑的色块中有星星点点的留白纸纹透出，边缘的轮廓线有轻微的油墨溢出与飞白。画面上下边缘有模拟老旧纸张的卷边与破损，以断续的轮廓线与网点交代。这种全方位的斑驳肌理让画面褪去了数字绘画的平滑，获得了手工版画或战时宣传海报的厚重与压抑，与“神话终结”的主题和哥特宗教的荒诞氛围完全契合。

Detalhes

GPT-Image-2

Replace the character in image1 with the character in image2, and maintain consistency in art style. Retain the original accessories of the character in image2, with clothing and body type referencing image1, and face shape referencing image2. Note that clothing colors should follow image2 but do not reference the clothing feature style of image2. Ensure no missing hair. Expression should reference image1, and do not include rings subtle luminous edges, flat shading, breaking the fourth wall, atmospheric, no visible grain, no pointillism, no dithering, 【Negative Prompt】 (low quality:1.4), (3d character), (realistic hair), (oil painting), (sketch), (bad lighting), (blending too much), (ugly)

Detalhes

GPT-Image-2

[Overall Style] vertical poster, semi-painted thick coating combined with letterpress printing texture, full color. The overall image uses sky blue, cream white, and orange red as the main tones, presenting a 1970 era printed poster style. The theme is a soda poster. The entire image is covered with a fine layer of grain noise, evoking a nostalgic atmosphere of the era. [Composition Design] In the center of the image is the figure of the 1 character from the legs upward, occupying the visual center. The upper part of the image features a sky blue diagonal strip, with white retro advertising font reading: “ Arctic Ocean Soda ”. In the lower right corner, in white advertising art font with a sky blue outline, it reads “ Arctic Ocean™ ” in small text. [Character Details] Positioned in the center of the image, leaning the body slightly and lifting one leg in an air-sitting pose, one hand holding a Arctic Ocean glass bottle filled with yellow soda (as shown in figure 2) raised to the cheek, the other hand giving a thumbs up and extending toward the camera. The character's eyes are bright, playfully closing one eye, with a sunny and confident smile. The gaze looks toward the camera. [Color and Texture] Cream white is the main background color. The entire image is covered with grain noise, with white and blue color blocks having a rough printed texture, fading the vividness, presenting a nostalgic atmosphere of the era.

Detalhes

GPT-Image-2

Horizontal illustration, retro mimeograph printing and minimalist flat vector style, full color, limited to a high-contrast palette of black, off-white, dark gray, and orange-red. This is an illustration of the space race era, radio fantasy, and a young girl wearing an astronaut helmet from Figure 1—silently transmitting and receiving between abstract geometric signals and the deep black cosmos. With Figure 1 as the protagonist, an asymmetrical left-right bipartite composition The image is clearly divided into two fields, left and right. The left side occupies about two-fifths of the frame, featuring a concrete half-length portrait of the girl, her form built from overlapping black and dark gray color blocks without contour lines, shaped solely by the boundaries of the color blocks. The right side occupies three-fifths, a vast expanse of deep black—not empty, but bearing a large abstract geometric figure that serves both as a cosmic background and a display screen for signals. The visual weight of both sides is balanced through precise area ratios: the “solid” on the left and the “void” on the right confront each other in silence. Left-side figure: Bear-eared girl wearing an astronaut helmet She faces right, her body in a half-profile pose, her gaze level, looking toward the deep black geometric field on the right side of the image. On her head, she wears a bulky Soviet-style astronaut helmet, painted flatly in off-white and light gray, with a highly prominent red star emblem on top and the Cyrillic letters “CCCP” printed on the front in bold, slightly worn-edged typeface. A pair of round bear ears naturally protrudes from both sides of the helmet, their soft curves contrasting with the helmet's hard sphere. She wears a black dress inside, a white apron over it, and a red scarf tied at the collar—a Soviet-style plain school uniform, creating a clear disjunction with the astronaut helmet symbolizing distance and technology. This state of “only wearing the helmet, not yet changed out of the school uniform” preserves her childish innocence and her most naive yearning for spaceflight. Her face is rendered as a flat off-white color block, with facial features outlined only in dark gray fine lines—a pair of round, quiet eyes, a small nose, lips slightly closed, an expression focused and slightly curious, without exaggerated emotion, more as if she is listening to something. Signal emission from the top of the head From the top of her helmet, several orange-red concentric arcs and short line segments radiate upward and outward, mimicking the abstract form of radio signals or Wi-Fi waves. These arcs have no gradients, all painted in solid colors, with deliberately retained rough jagged edges on the borders, simulating the misregistration texture of screen printing. They emanate from her, spreading upward and into the right space, naturally guiding the viewer's gaze toward the right half of the image. Right-side field: Radar and cosmic echoes In the lower right corner of the image is a large concentric semicircular wave pattern in orange-red, off-white, and black, resembling the echoes on a radar screen or the cross-section of a spinning record. The concentric semicircles rise from the lower right edge of the image, spreading outward ring by ring, with uneven spacing between the waves—dense near the center, gradually sparse outward, producing a slow, rhythmic visual vibration. To the upper left of this set of concentric waves floats a smaller radial signal pattern—four orange-red line segments radiating from a single center in diagonal directions, each ending with a short perpendicular tick, like an abstracted North Star or a simplified symbol of a rotating radio telescope. There is no connecting line between these two geometric forms, but they form a clear visual correspondence—one transmitting, one receiving, or both silently waiting for a response from farther away. The large expanse of deep black background envelops these shapes, making the orange-red and off-white appear particularly calm and powerful against the dark backdrop, like a faint but persistent signal source in deep space. Color system and mimeograph printing texture The entire image is strictly limited to four colors: black (cosmic background, dress, dark gray levels of shadows), off-white (helmet, apron, skin, some layers of the waves), dark gray (transitional color of clothing, midtones of geometric forms), and orange-red (red star, scarf, signal waves, radial symbols). No gradients, no highlights—all tonal transitions are directly cut by clean, sharp color block boundaries. The edges of the color blocks deliberately retain slight misalignments and irregularities—for example, at the junction of orange-red waves and the black background, extremely thin off-white gaps are left, mimicking the misregistration marks of old screen printing. Some color block surfaces have very slight dry brush effects and sparse dot patterns, as if the ink did not fully cover the paper, all reinforcing the hand-printed warmth of space race propaganda posters from the 1960s and 1970s. Overall atmosphere and emotional direction The image is quiet and restrained, without intense movement or dramatic narrative. She simply wears an adult's helmet and her own school uniform, facing the abstract waves and signals in a deep black cosmos. Whether those geometric patterns are echoes from the outside or questions emitted from her helmet need not be answered. The entire image lingers in this suspended moment—childhood and grandeur, school uniform and helmet, reception and transmission—all oppositions gently wrapped in flat color blocks, like an old poster smoothed by time, preserving an era's collective gaze at the starry sky in minimalist graphic language.

Detalhes

GPT-Image-2

ban2 竖版图像，画面比例2:3 生成一幅充满叙事张力的精美图片，使用例图中的角色。主体构图：画面主体是图一，二角色与图二汽车的合影画面。一.关于图一角色 1.动作：图一角色倚靠着图二汽车的车身站在街边的人行道上，重心压在车身上，手上拿着可乐，身体与车子形成稳定平衡的状态。转头看向镜头。图二角色从车中探出上半身带着笑脸向镜头招手，上半身在车窗外，另一只手撑在车门上体表细节：保留原图角色容貌画风特征。二.关于图二汽车：图二汽车型号为1959款凯迪拉克Eldorado。车身优美性感，漆面崭新完美表面锃亮无灰尘，街边景色的倒影映照在车身上。车停在路边，沿着路边整齐停放，不歪斜，远离街道中央。场景：场景在洛杉矶的一处富有地域气息的繁华街道旁，一个汽修店门口。时间在傍晚，天边晚霞灿烂。背景近处是来往的路人、车辆和街道。背景层次丰富，街景富有生活气息，现实感强烈。光影：柔和的侧逆光和环境光，晕影效果（vignette）。夕阳在角色身体轮廓和汽车车身轮廓形成锐利但柔和的暖金色光晕，角色身体、面部和汽车车身、车窗等形成大面积分割光和复杂立体的光影效果。高对比度照明，柔和但分明的大片深邃阴影，柔和过渡性的明暗对比。暖色调金橙色为主要色调。电影级光影效果，氛围感极强，使人身临其境。画面布局： 1.图一角色处于画面中心和焦点，近距离特写肖像。多展示图一角色的身体和面部细节。 2.图二汽车占据图一角色后方的画面，占据整个画面约二分之一的部分。部分车身处于画面外，部分车身被图一角色遮挡。 3.街景占据画面剩余部分。行人、其他车辆、街景和背景建筑全部虚化处理，模糊化以衬托画面主体的展示效果。所有元素共同将整个画面内容占满，画面中不留空白。人物、背景和场景共同融入画面中实现和谐统一的效果。画面层次分明，人物细节精细刻画，最近景和远景虚化，为画面增加丰富的层次感和精致感。镜头视角：平视视角，镜头进行极近特写拍摄，禁止俯视、仰视视角，禁止广角镜头。强烈透视畸变，压倒性视觉张力。画面氛围：图一角色以二次元动漫人物风格呈现，严格保留其一切二次元外貌画风特征，禁止出现写实人物风格。图二汽车和背景场景以真实照片风格呈现。图一角色完美融入背景中，不突兀不崩坏。怀旧氛围，色彩鲜艳，相机过曝的高饱和度画面，老照片颗粒感，光影质感精致细腻复杂，重点呈角色身上和车身上的光影效果。细节和注意事项：人物和车辆的大小对比符合现实比例。禁止出现人物体型过小的错误。图二汽车样式完全按照参考图和提示词生成，禁止出现任何车身扭曲、歪斜、比例错误等问题。禁止出现车身入侵挤占街道背景的情况。车身前后方向必须与人行道平行。技术要求：角色的服饰、发型、配饰等严格参照例图和提示词。画面完全忠实于提示。浅景深，背景虚化，杰作，最佳品质，鲜艳色调。请规避繁复的细微纹理、过度的细节刻画、颗粒噪点、粗糙质感、斜向网格纹路、瓷砖式重复纹样、泥污类叠加效果，同时去掉多余的闪光特效。弱化复杂纹理质感，优先保证画面表面顺滑均匀、轮廓线条粗实清晰、主体轮廓辨识度高。相较于堆砌细节，更注重造型利落、留白恰当、构图规整，打造干净的画面观感。正面提示词：高质量，高画质，无瑕疵，超精细细节。8K超清画质。贴合现实的精美细致细节，精美而立体复杂的光影效果，人物衣服要有贴合现实的质感和褶皱效果。负面提示词： (低质量:1.5), (最差质量:1.4), (模糊),(构图崩坏), (低分辨率), (3D渲染),(现实人物风格) ,(人偶质感), (人体结构错误), (偏离参考角色设计), (错误发色), (错误瞳色),(错误服饰), (多余肢体),(肢体扭曲),(肢体错位),(动作僵硬),(动作错误),(穿模),(表情崩坏),(比例崩坏),(偏离提示词),(单薄的光影)

Detalhes

GPT-Image-2

生成一幅充满生活叙事感的写实摄影级画面，使用例图中的手绘速写本与咖啡馆场景。画面主体是一只手举起一本绿色单色速写本，本子上的手绘内容与其身后真实咖啡馆场景形成"画中画"呼应。一.画面主体与动作 1.动作：画面下方伸入一只手，从本子底部托举，拇指压在封底右下角，其余四指藏在本子背后承重。手腕自然弯曲，小臂没入画面下缘不完整出现。本子被举到接近视平线的高度，摊开成一个完整跨页，书脊竖直居中，页面正对镜头不倾斜不歪扭。左右两页因装订自然形成轻微弧度，页面边缘微微翘起。本子左上角夹着一个深绿色金属长尾夹，夹住左页上缘。本子下方垂下一条深绿色缎带书签，自然下坠略有弯折。 2.主体细节：速写本为硬壳装订，封面与书脊呈深绿色，内页为米白色略带暖调的厚质纸。跨页上是全部用同一支深绿色针管笔完成的单色线描，无任何其他颜色，笔触为细密的手绘排线与交叉影线，非矢量非印刷感，保留手绘的轻微抖动与墨色深浅不匀。左页画的是咖啡馆吧台：吊灯一排悬垂、咖啡机、层架、瓶罐、吧台前两个短发人物侧影，人物为简笔无五官细节。右页画的是垂吊绿植与一丛棕榈状叶片植物，叶片以放射状短线排布。右页最右侧留出一条竖栏，自上而下画着四枚圆形与盾形的仿咖啡店印章图案，下方是手写英文"Coffee time"与日期"2025.7.11"，字体为手写体非印刷体。保留例图中所有手绘内容的构图与笔触特征。 3.场景：本子身后是一间真实的工业风咖啡馆，处于失焦状态。天花板悬挂大量小灯泡串灯，暖黄色光点密集散布。画面上方有一丛真实的垂吊绿色蕨类植物，叶片细长下垂，与本子上画的绿植形成呼应。中景可见木质吧台、深色天花板管线、木椅与木桌，右侧远处有零散的暖光吊灯。背景人物完全虚化不可辨识。 4.光影：室内暖光环境，主光来自天花板串灯，暖黄色调。速写本受光均匀，页面亮度高于背景，形成自然的视觉主体。背景暗部为深棕与墨绿色，与前景米白页面构成明暗对比。整体色调偏暖，绿色与暖黄为画面两大主色。 5.画面布局：速写本占据画面中部约百分之五十的面积，位于画面视觉中心偏下。手部与小臂占据画面下缘约百分之十五。背景咖啡馆填满其余空间，天花板绿植占据画面上缘。画面不留空白。浅景深，本子页面锐利清晰，背景强烈虚化形成柔和光斑（bokeh），串灯化为大小不一的圆形光点。层次为：手与本子（清晰前景）→ 咖啡馆中景（中度虚化）→ 远处吊灯（重度虚化）。 6.镜头视角：平视视角，镜头与本子等高正对拍摄，模拟手持随手记录的第一人称视角。中长焦定焦镜头质感，大光圈浅景深。禁止广角畸变，禁止俯拍或仰拍，禁止倾斜构图。 7.画风与氛围：整体为写实摄影风格，画面中的速写本内容保持纯手绘线描风格，两种质感在同一画面内并存且界限清晰。氛围安静、松弛、有生活气息，呈现旅行手账或城市speed记录的真实感。保留参考图的绿色单色线描画风特征与咖啡馆环境特征。 8.技术与细节要求：手部解剖结构准确，手指数量正确，关节弯曲自然，指甲形状真实，不得出现多指少指或关节反折。本子跨页的透视正确，书脊居中，左右页面对称不扭曲。纸张需有真实厚度与纤维质感，页面边缘可见分层。线描内容不得糊成一团，笔触需清晰可辨。规避过度锐化、HDR感、塑料质感、噪点堆砌。 9.正面提示词：大师级摄影作品，8K超清画质，超精细细节，真实纸张纤维质感，自然散景光斑，胶片级色彩过渡，高质量，无瑕疵。 10.负面提示词：(低质量:1.5),(最差质量:1.4),(模糊:1.2),(手部畸形:1.5),(多余手指:1.5),(手指粘连:1.4),(3D渲染),(塑料质感),(过度锐化),(HDR过曝),(文字乱码:1.3),(线描糊作一团),(书脊歪斜),(页面扭曲),(彩色线描),(印刷矢量感),(构图崩坏),(比例崩坏)

Detalhes

GPT-Image-2

生成一幅温暖怀旧的手绘漫画分镜画面，使用例图中的 Snoopy 与 Charlie Brown 角色形象与黄昏长椅场景。画面是一人一狗在夕阳下并肩坐在木长椅上的对话瞬间，情绪安静、略带惆怅。一.画面主体与动作 1.动作：画面左侧 Charlie Brown 侧身坐在木长椅左端，双腿自然下垂膝盖并拢，小腿略向内收，两只手放在膝盖上方，上半身微微前倾并转向右侧，头部转向右下方注视身旁的狗。画面右侧 Snoopy 端坐在同一条长椅的右半部，臀部落在椅面上，两条后腿在身前伸直贴着椅面，两只前爪并拢自然垂在身前，脊背挺直，头部微微上扬转向左侧回望 Charlie Brown。两者视线在画面中部相交，形成对角线呼应。 2.神态表情：Charlie Brown 眉毛下压成两道短弧，眼睛为两个小圆点，嘴角平直微微下撇，整体是若有所思的失落神情，不夸张不哭泣。Snoopy 单侧黑色大耳朵垂落，眼睛为一个黑色椭圆点，嘴部为经典的长吻线条，神情平静温和带一丝了然，不做笑脸不吐舌。 3.主体细节：Charlie Brown 为圆头光头造型，头顶只有一小撮弯曲的短发，身穿经典黄色短袖 T 恤，胸前横贯一道黑色锯齿状闪电条纹，下身深色短裤，脚穿棕色系带短靴，露出一截白色短袜。Snoopy 为白色圆润小猎犬造型，头部与身体为白色，单耳与背部有黑色块面，颈部系一条红色细项圈。严格保留例图中两个角色的经典造型比例与配色特征。 4.场景：长椅为深棕色横板木质结构，三到四条横向椅板，两侧为斜撑木腿，坐落在草地上。画面左侧后方矗立一棵粗壮大树，树干占据画面左缘，树皮为竖向纹理，茂密树冠从画面上缘垂下大片树叶，叶片为深浅不一的黄绿与橄榄绿色块。远景为开阔草坡与低矮灌木，地平线上方是黄昏天空。 5.光影：主光源为地平线附近的落日，位于画面右侧中下方，呈一个橙黄色半圆。光线自右向左低角度斜射，天空由画面右下的暖橙、金黄向左上过渡为紫灰与藕荷色，云层被染成橙红与紫灰条带。人物与长椅受右侧暖光照射，右侧轮廓偏亮，左侧落入柔和阴影，地面投下向左延伸的长影。整体色调温暖偏橙，暗部不死黑。 6.画面布局：竖构图。长椅与两个角色横贯画面中下部，占据画面高度约三分之一。Charlie Brown 位于画面左三分之一处，Snoopy 位于画面中央偏右。大树与树冠占据画面左侧与上缘约四分之一。画面右上方留出天空区域安放两个对话气泡，上方气泡略小位于中上部，下方气泡略大位于其右下方，气泡为米白色圆角矩形带指向说话者的尖角。落日位于画面右侧中下部靠近地平线。前景草地占据画面下缘。 7.镜头视角：平视视角，机位与坐姿角色视线基本等高，正面略偏侧的方向拍摄长椅。中焦镜头质感，全景深，前后景均清晰。禁止俯拍仰拍，禁止广角畸变，禁止倾斜构图。 8.画风与氛围：整体为经典美式报纸连环漫画风格，彩色铅笔与蜡笔质感的手绘上色，笔触有可见的颗粒感与排线，色块边缘柔和不锐利，轮廓线为深棕色手绘线条粗细不匀。氛围安静、温柔、带淡淡的乡愁。保留例图中的 Snoopy 与 Charlie Brown 角色特征与漫画画风特征，不做去 IP 化处理。 9.技术与细节要求：两个角色的头身比例须符合原作设定，Charlie Brown 为大圆头小身体，Snoopy 头部略大于身体。手脚数量正确不多不少，长椅木板透视一致不扭曲，椅腿着地稳固不悬空。对话气泡边缘平滑，尖角准确指向对应说话者。树叶不糊成一团色块需有层次。规避真实照片质感、3D 渲染质感、过度锐化、噪点堆砌。 10.正面提示词：经典美式连环漫画风格，彩色铅笔手绘质感，温暖黄昏光线，柔和色彩过渡，高质量，细节丰富，无瑕疵。负面提示词：(低质量:1.5),(最差质量:1.4),(模糊:1.2),(3D渲染:1.4),(真实照片:1.4),(比例崩坏:1.5),(多余肢体:1.5),(五官畸形:1.4),(文字乱码:1.3),(气泡尖角错位),(线条僵硬),(过度锐化),(色彩过饱和),(构图崩坏)

生成一幅柔和的手绘彩铅插画，主体是可达鸭（Psyduck）坐在浴缸的泡沫中洗澡，头顶顶着一只黄色小鸭子玩具，周围漂浮着透明肥皂泡。一.画面主体与动作 1.动作：可达鸭为正面偏左四分之三角度的坐姿，身体坐在一堆白色泡沫之中，臀部与下半身被泡沫完全淹没不可见。躯干略向画面右侧倾斜，头部保持正直并微微抬起。双臂自然下垂并向身体两侧微微张开，短小圆钝的手臂末端没入泡沫中，右臂（画面右侧）露出较多可见完整的圆弧轮廓，左臂（画面左侧）大部分被泡沫遮挡仅露出一小段。角色整体姿态放松、慵懒，呈现享受泡澡的松弛状态。头顶稳稳顶着一只黄色橡皮小鸭子玩具，玩具保持平衡不歪斜。 2.神态表情：呆滞、放空、略显困惑的招牌表情。左眼（画面左侧）为一只圆形的白色眼睛，中央有一颗小小的纯黑色圆瞳孔，瞳孔位置略偏内侧，眼神空洞无焦点。右眼（画面右侧）被一个大型的透明肥皂泡完全遮挡，透过泡泡可以隐约看见眼睛的轮廓与瞳孔，呈现折射后的轻微变形。嘴部为宽阔扁平的浅黄色鸭喙，喙尖朝向画面右侧，上下喙闭合，喙缝为一道向上微弯的细线，形成似笑非笑的憨态。整体神态传达出典型的迷茫、无所事事、什么都没在想的独特喜感。 3.主体细节：使用例图中的可达鸭角色设计特征，严格保留——身体为柔和的浅黄色（奶油黄），头部为圆润的椭圆形，与躯干直接相连无明显脖颈。头顶有三根细长的深棕色呆毛向上翘起并自然弯曲。鸭喙为比身体稍深的暖黄色，扁平宽阔，占据面部下半部约三分之一。身体表面用深棕色的细线勾勒轮廓，线条为手绘质感有轻微的粗细变化与断续。躯干正面有一道浅色的腹部分界弧线。头顶站立的小黄鸭玩具为经典橡皮鸭造型：圆胖的身体，橙色扁嘴，头顶有一小撮橙红色的呆毛，一侧有小小的翅膀凸起，通体为比可达鸭更饱和的柠檬黄色，同样以深棕色细线勾边。 4.场景：角色置身于一个浴缸内部，浴缸内堆满蓬松的白色肥皂泡沫。泡沫为大量圆形与椭圆形的白色团块相互堆叠而成，边缘用极淡的青灰色线条勾出层次，质地轻盈蓬松。泡沫层在画面下部三分之一处形成起伏的水平堆积，并在角色身体两侧向上堆高包裹住肩部。空中漂浮着七到八个大小不一的透明肥皂泡，最大的一个覆盖在角色右眼位置，其余分散在画面上方与右侧。每个泡泡以极细的浅蓝灰色圆环线勾勒，内部为半透明的浅蓝白色，泡泡上方有一小块白色月牙形高光，泡泡与背景之间可见轻微的透色关系。 5.光影：整体为柔和的散射光照明，无强烈方向性主光。光线来自画面上方偏左，角色头部与肩部上缘略亮，下半身接近泡沫处稍暗。可达鸭的黄色身体上有极轻微的暖黄色明暗过渡，靠彩铅的叠色深浅表现，不使用硬边阴影。泡沫的受光面为纯白色，背光的凹陷处为浅青灰色。肥皂泡上的白色高光为画面最亮点。整体明暗对比极低，属于柔和高调的照明，营造温暖朦胧的氛围。 6.画面布局：竖版构图。可达鸭占据画面中央约百分之五十的面积，位于画面中部略偏左，头部位于画面中上部，头顶的小黄鸭玩具接近画面上部三分之一处。白色泡沫层占据画面下部约三分之一并横向铺满。背景填充上部与两侧。角色身体下缘被泡沫自然遮挡，构图无生硬裁切。视觉重心落在可达鸭的头部与被泡泡遮住的右眼位置。 7.镜头视角：平视视角，镜头与角色头部大致等高正对拍摄，模拟蹲在浴缸边平视观看的角度。无明显透视变形，属于插画式的平缓视角。禁止俯拍，禁止仰拍，禁止广角畸变，禁止景深虚化。 8.画风与氛围：手绘彩色铅笔插画风格，可见清晰的彩铅笔触与纸张颗粒纹理，色彩通过多层轻薄叠加获得而非平涂。轮廓线为深棕色的手绘细线，有自然的粗细变化与轻微抖动，转角处不完全闭合留有呼吸感。配色为极低饱和的柔和粉彩调：奶油黄、柠檬黄、浅粉、淡青蓝、纯白、深棕。背景上部为柔和的浅粉色晕染，下部过渡到淡青蓝色，两色以柔和的水平分界带相接，分界处有轻微的笔触混色。背景上散布若干极淡的白色小圆点作为纹理装饰。整幅画面覆盖一层温柔的纸张颗粒感与轻微的褪色老照片质地。氛围温暖、治愈、慵懒、童趣。保留例图中的角色造型比例、配色方案与识别特征。 9.技术与细节要求：可达鸭的身体比例需符合原作设定——头部约占全身高度的一半，鸭喙宽度约为头宽的三分之二，头顶呆毛为三根不得增减。双眼中可见的那只眼睛需为规整的圆形，瞳孔为小黑点不得画成大瞳孔或有复杂高光。手臂末端为圆钝的无指形态，不得画出手指。头顶小黄鸭玩具必须稳定站立在头顶正中偏左位置，与头部有明确的接触关系不得漂浮。肥皂泡必须为规整的正圆形，透明度一致，遮挡眼部的那个泡泡需正确呈现透视折射关系而非完全不透明的白球。泡沫团块之间需有层次遮挡关系，不得糊成一片均匀白色。彩铅笔触方向需有变化不得全部平行机械排列。不得出现多余的肢体、重复的角色或漂浮不合理的元素。 10.正面提示词：手绘彩色铅笔插画，柔和粉彩配色，可见彩铅笔触与纸张颗粒，深棕色手绘勾边，柔和散射光，透明肥皂泡质感，蓬松泡沫层次，温暖治愈氛围，高质量，无瑕疵。 11.负面提示词：(低质量:1.5),(最差质量:1.4),(模糊:1.3),(3D渲染:1.5),(写实照片:1.4),(数码平涂:1.3),(硬边阴影:1.3),(高饱和:1.4),(角色形象偏离原作:1.5),(错误配色:1.4),(身体比例失调:1.4),(多余肢体:1.4),(画出手指:1.3),(泡泡不透明:1.3),(泡沫糊成一片:1.3),(矢量均匀线条:1.3),(过度锐化),(HDR),(水印),(文字),(签名),(边框)

Detalhes

How it works

Veja a imagem e o prompt

Cada card mostra uma imagem publicada do GPT-Image-2 ao lado do prompt completo utilizado.

Carregue no gerador

Ao escolher «Gerar», o prompt entra sozinho e o modelo passa para GPT-Image-2. O texto puro também pode ser copiado.

Mude uma variável de cada vez

Rode primeiro o prompt original e depois altere uma única exigência, para saber o efeito de cada mudança.

About GPT-Image-2 and Prompts

Esta biblioteca reúne gerações do GPT-Image-2 publicadas no Labnana junto de seus prompts. Cada imagem corresponde ao texto que realmente a produziu; não há exemplos escritos sem verificação.

O GPT-Image-2 é o modelo de geração de imagens da OpenAI. No Labnana o mesmo prompt também pode rodar na família Nano Banana do Google ou no Seedream 5.0 Pro da ByteDance, para comparar resultados.

Gerar com o GPT-Image-2

Other models on Labnana

All models

GoogleNano Banana PromptsIn-image text, character consistency, photoreal detail

ByteDanceSeedream 5.0 Pro PromptsRegion-level edits, local corrections, Chinese instructions

Frequently asked questions

Comece com o GPT-Image-2

Carregue qualquer prompt desta página ou escreva o seu.

Ver preços