How to write GPT Image 2 prompts for posters, banners, slides and product images
For campaign ready images, write the prompt like a design brief: purpose, subject, composition, style, text and constraints. The most important details for marketing assets are aspect ratio, subject placement, negative space, exact on image text and limits such as no extra text, watermark or unrelated logos.
Published byEdited with GPT-5.5Images generated with GPT Image 2
For campaign ready images, write the prompt like a design brief: purpose, subject, composition, style, text and constraints.
The most important details for marketing assets are aspect ratio, subject placement, negative space, exact on image text and limits such as no extra text, watermark or unrelated logos.
When editing or using reference images, state what must stay unchanged, what may change and use narrow instructions such as change only the background.
Prompt GPT Image 2: công thức brief thiết kế cho poster, banner, slide, ảnh sản phẩmMinh họa AI cho công thức prompt dạng brief thiết kế.
AI Prompt
Create a landscape editorial hero image for this Studio Global article: Prompt GPT Image 2: công thức brief thiết kế cho poster, banner, slide, ảnh sản phẩm. Article summary: Để GPT Image 2 tạo đúng poster, banner, slide hoặc ảnh sản phẩm, hãy viết prompt như một brief thiết kế 6 phần: mục đích, chủ thể, bố cục, phong cách, text và ràng buộc; nếu dùng API, cần kiểm tra đúng tên model vì tà.... Topic tags: ai, openai, chatgpt, image generation, prompt engineering. Reference image context from search candidates: Reference image 1: visual subject "Nếu bạn đang làm marketing và phải liên tục tạo hình ảnh cho quảng cáo, social post, landing page... thì việc biết cách viết prompt tạo ảnh AI sẽ giúp bạn tiết kiệm hàng giờ – và c" source context "Prompt tạo ảnh AI: Vũ khí mới cho marketer 2026" Reference image 2: visual subject "Nếu bạn đang làm marketing và phải liên tục tạo hình ảnh cho quảng
openai.com
If you want GPT Image 2 to create something you can actually use in a campaign, presentation deck or product page, do not start with a vague request like make a nice poster. A better prompt reads like a short design brief: what the image is for, what the main visual is, how the layout should work, which words must appear exactly, and what the model must not add or change.
OpenAI’s image generation and image prompting guidance repeatedly points in this direction: describe the purpose, composition, style, text and constraints clearly; for more complex requests, split the prompt into labelled sections so the model has a structure to follow.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
What is the short answer to "How to write GPT Image 2 prompts for posters, banners, slides and product images"?
For campaign ready images, write the prompt like a design brief: purpose, subject, composition, style, text and constraints.
What are the key points to validate first?
For campaign ready images, write the prompt like a design brief: purpose, subject, composition, style, text and constraints. The most important details for marketing assets are aspect ratio, subject placement, negative space, exact on image text and limits such as no extra text, watermark or unrelated logos.
What should I do next in practice?
When editing or using reference images, state what must stay unchanged, what may change and use narrow instructions such as change only the background.
A note on naming: this article uses “GPT Image 2” because that is how many people search for the workflow. In actual use, especially in the API, check the current model name in the product or documentation you are using. The OpenAI sources cited here refer to GPT Image Generation Models, gpt-image-1.5 in the API, ChatGPT Images and ChatGPT Images 2.0 in different contexts.
The six-part formula: write the prompt like a design brief
This is not an official OpenAI formula. It is a practical way to organize the recommendations into a reusable prompt structure. The goal is to lock down the parts that usually cause image results to drift: purpose, layout, text and constraints.
Prompt part
What to specify
Quick example
Purpose
Whether the asset is a poster, banner, slide, social ad, product image or infographic
Vertical poster for an AI workshop
Main subject
The product, person, setting, object or action at the centre of the image
A serum bottle on a marble surface
Composition
Aspect ratio, camera angle, framing, subject placement and empty space for copy
16:9, product on the right, clean space on the left
Style
Realism level, materials, lighting, colour palette and mood
Studio product photo, soft light, premium beige palette
Text in the image
Exact wording, type style, colour, relative size and placement
Text: ‘SUMMER SALE’, bold white sans-serif, top left
Constraints
What must not be added, or what must stay unchanged
No extra text, no watermark, no unrelated logo, preserve product geometry
Constraints such as no trademarks, no watermarks, no logos, include only this text verbatim and do not redesign the packaging appear in OpenAI’s prompting guidance for situations where design control and packaging accuracy matter.
A reusable prompt template
When a prompt has several requirements, use labelled lines instead of cramming everything into one long sentence. That structure makes it easier to check whether you have covered each part of the brief and matches OpenAI’s guidance for complex image prompts.
Create a [poster/banner/slide/product image] for [purpose].
Subject: [main subject].
Composition: [aspect ratio, placement, negative space, camera angle].
Style: [visual style, lighting, colour palette, realism level].
Text: '[exact text]', [font/style/size/colour/placement].
Constraints: [no extra text / no watermark / no unrelated logos / preserve layout / preserve product geometry].
Four ready-to-adapt prompt examples
Use these as starting points. Replace the bracketed sections with your real campaign, product or presentation details.
1. Event or campaign poster
Create a vertical 4:5 poster for [event or campaign].
Purpose: promote [goal].
Subject: [describe the main visual].
Composition: main subject centred, with clear negative space at the top for the headline.
Style: [editorial / cinematic / minimal], [lighting description], [main colour palette].
Text: show only '[HEADLINE]', in bold white sans-serif, placed at the top of the poster.
Constraints: no extra text, no watermark, no unrelated logos.
2. Website or landing-page banner
Create a wide 3:1 website banner about [topic].
Subject: place the main subject on the right; keep the left side clean for copy.
Style: modern, premium, uncluttered, [colour palette], soft lighting.
Text: show only '[TEXT]', placed on the left, high contrast and easy to read.
Constraints: no extra text, no watermark, no unrelated logos.
3. Presentation slide background
Create a 16:9 slide background for a presentation about [topic].
Style: clean, professional, low detail, with enough contrast for overlaid text.
Composition: leave a large empty area on the left for a title and three bullet points; place the main visual on the right.
If text appears in the image, use only '[TEXT]' and make it sharp.
Constraints: do not add icons, labels, logos or watermarks unless requested.
4. Commercial product image
Create a photorealistic product image of [product name] on a [colour or material] background.
Composition: 3/4 front view, soft studio lighting, realistic reflections and natural shadow.
Packaging: the product label should be crisp and readable: '[TEXT]'.
Constraints: if a reference image is provided, preserve the product geometry; do not redesign the packaging; do not add unrelated objects, watermarks, logos or trademarks.
How to control text inside the image
If the asset needs a headline, product label, call to action or slide title, make the text its own prompt section. OpenAI recommends writing the text verbatim, specifying placement and formatting, and keeping text short when possible; for text-heavy layouts such as infographics, you should still review the output carefully and expect to refine it in a design tool.
Write the exact words: for example, Text: 'LAUNCH WEEK'.
Keep the copy short: a headline, label or short CTA is easier to inspect than a paragraph.
Specify type style: bold sans-serif, elegant serif, monospaced or handwritten.
Specify placement and colour: large white text at top left; small black text under the product label.
Block stray text: add only include this text or no extra text.
For names or difficult wording: spelling the phrase character by character can help the model understand the required text.
When using reference images or editing an existing image
If you upload a product photo, layout or style reference, tell the model what role each image plays. OpenAI’s material on image generation and editing stresses the value of describing what should be preserved, what should change and which constraints prevent unwanted edits.
Image 1 is the original product image. Image 2 is the style reference.
Keep the shape, proportions, label and layout from Image 1.
Apply only the lighting, colour palette and mood from Image 2.
Do not redesign the packaging. Do not add new text, watermarks or logos.
For small edits, make the request narrow and explicit:
Change only the background colour to warm beige.
Keep the product shape, label placement, lighting, shadows, camera angle and composition exactly the same.
No extra text, no watermark.
For product and packaging work, constraints such as preserve product geometry, do not redesign the packaging and keep label placement unchanged are especially useful when shape, labelling and brand presentation need to remain stable.
How to fix an image that is close but not quite right
If the result is close, do not automatically rewrite the whole prompt. OpenAI recommends improving results by adjusting specific elements, because changing too much at once can make the image drift away from a composition or style that was already working.
Too dark: add make the lighting brighter and softer.
Too cluttered: add simplify the background or increase negative space.
Right layout, wrong style: write keep the same composition, make the style more modern.
Wrong text: reduce the number of words, write the exact text again, add only include this text, then inspect the new output carefully.
Distorted product or changed packaging: add preserve product geometry, do not redesign the packaging and keep label placement unchanged.
Pre-flight checklist before generating
Before you send the prompt, quickly check:
Is the asset a poster, banner, slide background or product image?
What aspect ratio do you need: 4:5, 16:9, 3:1 or something else?
Where should the main subject sit in the frame?
Do you need negative space for copy or later layout work?
Which text must appear exactly as written?
Have you specified the type style, colour, relative size and placement?
Have you blocked extra text, watermarks, unrelated logos or strange objects?
If using reference images, have you said which image is the product and which is the style reference?
For product images, have you asked the model to preserve geometry, label placement and packaging?
The bottom line: a strong GPT Image 2 prompt is not necessarily the longest prompt. It is the clearest brief. Define the purpose, subject, composition, style, text and constraints, and you are more likely to get a poster, banner, slide or product image that matches the idea you started with.
