Prompts

GPT Image 2 提示词

上方保留现成生成器,下方放一组精选模板。想更快起步时,直接点一键试用,把提示词回填到上面的生成器里再继续微调。

GPT Image 2——AI 图像生成器与编辑器
适合海报、产品视觉、可读文字版式和参考图编辑。登录后开始生成,当前每次生成消耗 3 积分。
必填
0 / 20000
3 积分 / 本轮(1 张)登录后开始生成
生成结果
任务完成后,生成图片会自动显示在这里。

你的 GPT Image 2 结果会显示在这里。

生成完成后,结果区会自动刷新,方便你直接预览与下载。

精选模板

GPT Image 2 提示词库,直接服务生成器

每张卡片都是一个真实可用的起点。点击一键试用后,提示词会回到上方生成器,你可以继续改几个词再生成。

33 条精选提示词

Create a cohesive strawberry matcha campaign for Kizuki, a newly opened matcha shop in Brooklyn Heights. Generate four coordinated deliverables in one unified visual language: a Twitter post, an Instagram Story, an Instagram feed post, and a LinkedIn format. Keep the hero drink, styling, props, lighting, and brand mood consistent across all four sizes while adapting the layout naturally to each platform.

Alternate-history high school computer lab in 2002 where every student is using ChatGPT. Beige CRT monitors, Windows XP browser windows, ball mice, tangled cables, backpacks on the floor, fluorescent classroom lighting, and a believable early-2000s school atmosphere. Add a subtle orange date stamp in the lower-left corner that reads "02 18 04".

A 35mm film photography book page featuring 1970s New York street photography. Authentic grain, candid sidewalks, imperfect framing, warm paper tone, and the feeling of flipping through a serious photo monograph.

A dramatic page from a Japanese seinen manga, with mature atmosphere, cinematic paneling, strong black ink shapes, expressive faces, and grounded tension. Make it feel like a printed manga page rather than a digital mockup.

A whimsical sequential comic about a tiny pastel-colored car, rendered in the style of a medieval illuminated manuscript. Flat decorative perspective, hand-painted textures, ornamental borders, and a playful storybook tone.

Based on the uploaded real portrait, create a manga-style character sheet that transforms the subject into a shonen anime character named Adele. Include a clean character turnaround, facial expressions, outfit details, pose studies, and polished notes that make it feel like a professional character-design presentation.

A serious creative studio desk covered with GPT Image 2 launch materials: review sheets, pinned wall proofs, print samples, layout studies, handwritten notes, crop marks, variations, and other pre-release design collateral. Make it feel like a meticulous art-direction workspace preparing for a major launch.

A 3:1 ultrawide action breakdown of a basketball dunk, showing the full sequence of motion across the frame from takeoff to finish. Keep the athlete readable at every phase, use crisp sports photography energy, and make the progression feel analytical and dynamic at the same time.

An iPhone panorama of a dense Thai city scene, captured with authentic handheld energy and a slight intentional stitching misalignment. Keep the light, color, and street detail realistic, as if it came straight from a phone camera roll.

An iPhone panorama of summer in the South of France, with bright sun, holiday color, relaxed coastal atmosphere, and believable handheld phone-camera realism. Let the scene stretch naturally across the frame like a real panoramic sweep.

An infographic that explains Cantor's diagonal argument by visualizing the proof in four clear stages: assumption, diagonal, construction, and contradiction. Make the mathematics easy to follow at a glance, with restrained design, strong hierarchy, and precise step-by-step visual logic.

Using the uploaded research paper PDF as the source, turn the content into a polished academic poster. Preserve the paper's core argument, methods, results, and conclusions, while translating them into a clean, readable conference-poster layout with clear visual hierarchy.

Based on the uploaded portrait, create a personal color analysis and palette-matching diagnosis with minimal text. Show flattering swatches, a clear color direction, and a refined beauty-editorial presentation that feels clean, personal, and easy to understand.

A cinematic aerial view of a row of pink coastal motels, each with a differently shaped swimming pool, designed like a polished creative-demo travel image. The motel signs should read "Firefly Motel", "Firefly Lodge", "Firefly Stay", and "Casa Firefly". Keep the composition sunlit, graphic, and irresistibly vacation-like.

For OpenArt using GPT Image 2 as the creative director for Smart Shot movie-grade video production, create a cinematic storyboard sequence of an epic battle between Zeus and Hades. Frame it like a high-end previsualization deck with strong shot variety, mythic scale, and clear action continuity.

A 2015 lecture hall at the University of British Columbia where a professor is teaching GPT imagegen 2, and the projected slide shows that same professor teaching GPT imagegen 2 again in an infinite recursive loop. Keep the academic setting believable and make the recursion immediately legible.

A well-designed map guidebook page that combines cartographic clarity, editorial layout, labels, legends, and travel-handbook usability. Make it feel like a printed guide someone would actually carry.

A clear step-by-step visual guide to tying shoelaces. Show each stage cleanly, with intuitive sequencing, simple diagrams or illustrations, and a practical instructional design style.

A step-by-step instructional layout that teaches how to fold a paper crane. Make the process easy to follow, with crisp stages, clean visual direction, and a calm handmade feel.

Create a sample page of a dramatic Japanese manga in which the protagonist discovers a magical feather pen called the Quill of GPT Image. The entire page should be in Japanese, with strong panel composition and the feeling of a photographed printed comic page rather than a digital layout.

A bookstore display in India featuring book covers in Hindi, Bengali, Marathi, Telugu, Tamil, Urdu, Gujarati, Kannada, and Odia. All cover text should be sharp and clearly readable, and every book should show the same publisher mark: OpenAI. Arrange the display like a real bookstore shelf with strong typographic variety.

A Korean-language booking card for a hanok hotel, structured as three connected scenes within one polished ad layout. Keep the Korean title crisp and legible, and make the design feel premium, calm, and travel-ready.

A four-page comic about a capybara and an otter going on holiday in the South of France. Keep the two main characters visually consistent across all four pages while letting the locations, gags, and vacation scenes change around them.

A glossy science-magazine editorial page titled "North American Wolves Are Far Less Dangerous Than Most People Imagine". Smooth, restrained layout, polished magazine pacing, understated supporting visuals, and an authoritative but accessible science-publication feel.

The history of baseball in Toronto handwritten in pencil on 8.5×11 lined paper. Let the strokes vary naturally like a real person's writing, keep the page slightly imperfect, and add a faint coffee stain in the upper-right corner.

Create a 4:5 cover poster for this release with the theme "visual polyglot". Put the title "Create Everything at Once" in the center. Pack the composition with scientific charts, the periodic table, the solar system, medieval manuscripts, botanical illustrations, anatomical drawings, old maps, climate charts, engineering diagrams, transit signage, comic panels, UI screenshots, butterfly specimens, pie charts, and architectural blueprints, all layered into one ambitious editorial collage.

以手机随手拍摄的日常风格,桌面上摆放着一款埃隆马斯克动漫手办,动作夸张帅气,装备齐全。与此同时,真实世界中的埃隆马斯克也出现在镜头中,与手办摆出相似姿势,形成手办与真实人物同框的有趣对比。背景浅淡柔和并带有自然虚化,两位主体清晰突出,整体构图和谐自然,传递温暖而有生活气息的视觉体验。

A page of a comic book in the style of a modern indie comic, telling a story for 6-year-old kids.

1960s French New Wave theatrical poster, bold photomontage composition, torn-paper collage sensibility, pop-art color bursts, high-contrast black-and-white imagery with selective red blue and yellow accents, hand-made offset-print texture, slightly off-register ink, expressive asymmetry, art-house poster cool, graphic spontaneity, street-poster energy, adventurous typography-led design. Poster text: Large title at the bottom: "GPT Image 2.0". Smaller headline at the top: "Image generation with a point of view". Small footer text: "Coming soon". Keep all visible text in English. Use a theatrical poster composition.

Make a sample page of a colorized Japanese shonen adventure manga. The page should vividly depict our main character finding a magical quill called the Quill of GPT Image. Make it dramatic. The magical quill has strong power sealed inside it. Additional instructions: Aspect ratio portrait 1440x2560. The pen should have an OpenAI logo on it. The language throughout the manga should be Japanese. Think carefully first to make this a good story with a strong split of manga panels. The page should appear as a photo of a physical page, not a digital page.

I am creating a magazine page with the theme of "visual polyglot". The title in the center of the image should be "Create Everything at Once". Create a piece of art celebrating visual creations across the full breadth of human visual culture and natural visual elements. Include a curated collage of scientific diagrams, the periodic table, the solar system, medieval manuscript pages, botanical illustrations, anatomical drawings, old maps, climate charts, engineering schematics, transit signage, multilingual text, comic panels, UI screenshots, a camera photo, a butterfly specimen, pie charts, architectural blueprints, and facade drawings. Also think beyond these examples with other elements and styles that fit a premium research announcement or museum-style manifesto. Use an unstructured creative layout such as a fan-out, avoid grids, use a portrait 4:5 aspect ratio, do not add any content text besides the title, and avoid an overall beige tint so vibrant elements stay vibrant.

Hyper-realistic digital illustration of the historic Shanghai Bund waterfront looking across the Huangpu River to the futuristic Lujiazui district, presented as a single continuous composition showcasing the cycle of seasons. The scene flows seamlessly from left to right in a natural progression: winter, spring, summer, and autumn. The left side features cold snowy winter elements, gradually thawing into fresh green buds and blooms of spring, then morphing into lush vibrant vegetation and bright sunlight of summer, and finally transitioning into the golden orange and red hues of autumn on the far right. There are no visible dividing lines between seasons; the weather, lighting, and vegetation blend smoothly to create a unified and harmonious panorama. Rich in detail, symbolic of the passage of time, cinematic lighting, 8k resolution, highly detailed textures, aspect ratio 4:3.

这页是什么

什么是 GPT Image 2 提示词?

GPT Image 2 提示词,本质上就是告诉模型“你要生成什么、画面应该长什么样、气质应该偏什么方向、最终要服务什么用途”的自然语言指令。好的提示词通常会把主体、场景、光线、风格、构图和使用场景放进同一个请求里,让模型更容易理解你真正要的商业画面。

所以这页并不是为了做一个很重的数据库,而是提供一组真正可用的起点。你可以把它理解成“已经帮你写过第一稿的提示词库”。当你想做海报、电商商品图、发布主视觉、创作者封面或者参考图重绘时,不用从空白输入框开始,而是先拿一个比较接近目标的模板,再按自己的品牌和场景去改几个关键词。

这样的页面结构对 SEO 也更友好。搜索 GPT Image 2 prompts 的用户,通常不是只想看概念解释,他们更想看到可直接试用的例子、简洁的说明、以及一个马上能操作的生成器。上面是工具,下面是模板,再往下是 FAQ,这样的结构会更完整,也更像一个真实有效的产品页。

怎么用

怎么在这个页面上使用 GPT Image 2 提示词

这页的流程故意保持简单,核心是让用户少想一步,直接开始试。

先选一个最接近目标的模板

不要一开始就追求完全定制,而是先选一个最接近你任务的模板,比如海报、商品主图、发布会视觉或者创作者封面。模板越接近目标,后面需要改的内容就越少。

一键回填到上方生成器

点击一键试用后,提示词会自动回到顶部生成器。然后你可以把品牌名、产品类型、场景、风格或者传播用途改成自己的真实需求,而不是直接用一个完全泛化的模板去生成。

生成之后继续微调

第一轮生成的意义,更多是确认方向对不对。通常只要继续微调光线、镜头语言、文字空间、色调或者材质描述,就能把一个通用模板逐步推进成更适合实际投放或展示的画面。

FAQ

GPT Image 2 提示词 FAQ

把用户最常问的几个问题放在这里,方便补足页面结构,也方便搜索理解主题。

什么样的 GPT Image 2 提示词算好?

好的提示词不是越长越好,而是越清晰越好。通常一个有效提示词会明确主体、场景、光线、风格、构图重点和最终用途,让模型知道这是一张海报、商品图、封面还是营销主视觉。

提示词应该写短一点还是长一点?

没有固定答案。方向明确、允许模型自由发挥时,短提示词就够了;当你需要控制画面层级、商业质感、标题留白、产品呈现方式时,适度更长的提示词会更稳定。关键不在字数,而在信息是否清楚。

这些模板只能做文生图吗?

不一定。很多模板同样可以迁移到参考图重绘场景里使用。做法是保留模板里的商业方向和风格要求,再补一句“基于参考图进行哪些调整”,这样就能把模板转成 image-to-image 的指令。

为什么已经有生成器了,还要做 prompts 页面?

因为很多用户真正卡住的不是“不会点生成”,而是“不知道第一句怎么写”。提示词库能明显降低空白输入框带来的阻力,也能让页面更符合搜索意图,让用户从搜索结果进入后马上就能开始尝试。