Graphic Design · AI imagery
The Best Prompt for AI Image Generation
The best prompt for image generation is one that makes a text model do the describing for you. You state the subject, the purpose, the style and the tool; this prompt has the model set out the visual decisions (composition, lighting, color, medium, level of detail) and then write three ready-to-paste prompts that vary one decision at a time. It also returns a list of things to exclude and notes on what to change if the first results miss. The text model writes the prompts; a separate image tool makes the pictures.
Image Prompt Writer
A prompt that writes prompts: describe the image you need and get three detailed, copy-ready image-generator prompts plus notes for iterating.
- Works with
- ChatGPT, Claude, Gemini
- Difficulty
- Beginner
- Complexity
- Multi-part reply
- Variables
- 6
- Version
- 1.0
Customize this prompt
Fill in what you know. Anything left blank stays as a [bracketed placeholder]. Nothing you type leaves your browser.
6 variables · fields marked * matter most
You write prompts for AI image generators. Turn the request below into precise visual descriptions an image model can act on. You are writing text, not creating images. IMAGE REQUEST Subject: [Subject] What the image is for: [What the image is for] Style or medium: [Style or medium] Image tool I will use: [Image tool] Format: [Format] Must include or must avoid: [Must include or must avoid] STEP 1 - VISUAL DECISIONS Before writing any prompt, settle these and show them as a short list: - Subject: exactly what is shown, with the concrete details that matter (materials, age, clothing, number of items). - Setting and background. - Composition: framing, viewpoint, where the subject sits, and where empty space falls. Compose for the purpose, for example room for a headline. - Lighting: source, direction, quality, time of day. - Color: dominant colors and overall mood. - Medium and finish: the characteristics of the medium (lens and depth of field for photography; line, texture and shading for illustration). - Level of detail: what is sharp and what is left simple. Where my request leaves a decision open, make it and mark it [CHOSEN]. If the request contradicts itself, say so before continuing. STEP 2 - THREE PROMPTS Write three prompts, each one copy-ready paragraph of 50 to 110 words in natural descriptive language, starting with the subject. - Prompt A: the most direct interpretation of my request. - Prompt B: the same subject and style with a different composition. - Prompt C: the same subject and composition as A with different lighting and color mood. Beneath each, add one line saying what changed and when to pick it. STEP 3 - EXCLUSIONS AND ITERATION - Avoid list: elements to keep out, as a short comma-separated list. - If it goes wrong: four likely problems with the results and the wording change to try for each. RULES - Describe what is visible. Replace abstract words such as "innovative" with things a camera or illustrator could show. - Describe style by its properties: medium, palette, line, texture, era. Do not name living artists, studios or copyrighted characters. - No real, identifiable people and no brand logos or trademarks unless I state that I hold the rights. - If the image needs text, keep it to a few words in quotation marks and warn me that lettering often renders wrongly. - Do not invent tool-specific parameters or syntax. Use plain language unless I have given you the tool's syntax. State the format in words. - No generic quality terms such as "masterpiece"; specific description does more. OUTPUT Use the headings Visual decisions, Prompts, Avoid list, If it goes wrong. Put each prompt in its own code block.
Use it when you are
- Creating a hero or header image for a website, article or newsletter
- Producing concept visuals or mood images for a pitch, a storyboard or a client presentation
- Generating consistent illustration prompts for a set of social media posts
- Exploring art direction options before a photo shoot or an illustration commission
- Improving a short image prompt that keeps giving generic results
Why this prompt works
It makes the art-direction decisions before the wording
Short image prompts leave composition, lighting and color to chance, which is why results look generic. Step 1 has the model decide each of those in the open, in the order a photographer or illustrator would consider them, and mark its own choices. You can overrule a decision in one line before any prompt is written around it.
Purpose shapes the frame
An image for a website hero needs empty space for a headline; an avatar needs a centered subject that reads when small. Asking what the image is for, and telling the model to compose for it, addresses a very common complaint about generated images: a good picture that does not fit the layout it was made for.
Three prompts that vary one thing each
When variations change everything at once you cannot tell which change helped. Here B changes only composition and C changes only lighting and color, both measured against A. Comparing the results tells you which lever matters for this image, and the next round can combine the better composition with the better light.
Style is described, not borrowed
Naming an artist is a shortcut that raises ethical and legal questions, and some tools refuse it. Describing the medium, palette, line quality and era gives the image model more precise direction and keeps the result your own art direction. The same rule covers real people and trademarks.
No invented syntax
Image tools differ in how they take aspect ratios, exclusions and other settings, and those conventions change. A text model asked to add parameters may produce ones that do not exist. The prompt keeps everything in plain language unless you supply the syntax yourself, so the output works as a description in any tool.
How to use it
- Describe the subject as a scene: who or what, doing what, where. One or two sentences is enough.
- State the purpose and the format. Mention where text or interface elements will sit over the image.
- Pick a style, or choose "Let the model propose one" and judge its suggestion in the visual decisions.
- Check the Step 1 list before reading the prompts. If a [CHOSEN] decision is wrong, correct it and ask for the prompts again.
- Paste Prompt A into your image tool and set the aspect ratio in the tool's own settings. Generate a few images, then try B and C.
- When a result is close, use the "If it goes wrong" notes to adjust a single phrase at a time, and keep the wording that worked for later images in the same set.
AI model compatibility
| Model | Status | Notes |
|---|---|---|
| ChatGPT | Expected to work | If image generation is available in your plan, you can ask for one of the three prompts to be rendered in the same conversation, then quote the "If it goes wrong" fix that applies and try again. |
| Claude | Expected to work | If you have a style guide or a description of earlier images to match, paste it inside <style_reference> tags below the request; Anthropic recommends XML-style tags for separating reference material from instructions. |
| Gemini | Expected to work | Keep the step order. Google's prompt design guidance favors explicit, sequenced instructions, and Step 1 is what keeps the three prompts consistent with each other. |
“Expected to work” means the prompt uses plain, model-independent structure and has been reviewed for that — not that it has been run on that model. A model is marked Tested only when a test record is published below.
Worked example
Scenario: A small pottery studio needs a hero image for its new website and has no budget for a photo shoot yet.
You write prompts for AI image generators. Turn the request below into precise visual descriptions an image model can act on. You are writing text, not creating images. IMAGE REQUEST Subject: A ceramicist in her forties shaping a stoneware bowl on a wheel in a small, tidy studio, hands and forearms covered in wet clay What the image is for: Website hero image with a headline and a button placed over the left third Style or medium: Photorealistic photograph Image tool I will use: The image generator built into my chat app; no special syntax Format: Wide landscape (16:9) Must include or must avoid: Include soft daylight from a window. Avoid visible text, logos, cluttered shelves and an over-saturated look STEP 1 - VISUAL DECISIONS Before writing any prompt, settle these and show them as a short list: - Subject: exactly what is shown, with the concrete details that matter (materials, age, clothing, number of items). - Setting and background. - Composition: framing, viewpoint, where the subject sits, and where empty space falls. Compose for the purpose, for example room for a headline. - Lighting: source, direction, quality, time of day. - Color: dominant colors and overall mood. - Medium and finish: the characteristics of the medium (lens and depth of field for photography; line, texture and shading for illustration). - Level of detail: what is sharp and what is left simple. Where my request leaves a decision open, make it and mark it [CHOSEN]. If the request contradicts itself, say so before continuing. STEP 2 - THREE PROMPTS Write three prompts, each one copy-ready paragraph of 50 to 110 words in natural descriptive language, starting with the subject. - Prompt A: the most direct interpretation of my request. - Prompt B: the same subject and style with a different composition. - Prompt C: the same subject and composition as A with different lighting and color mood. Beneath each, add one line saying what changed and when to pick it. STEP 3 - EXCLUSIONS AND ITERATION - Avoid list: elements to keep out, as a short comma-separated list. - If it goes wrong: four likely problems with the results and the wording change to try for each. RULES - Describe what is visible. Replace abstract words such as "innovative" with things a camera or illustrator could show. - Describe style by its properties: medium, palette, line, texture, era. Do not name living artists, studios or copyrighted characters. - No real, identifiable people and no brand logos or trademarks unless I state that I hold the rights. - If the image needs text, keep it to a few words in quotation marks and warn me that lettering often renders wrongly. - Do not invent tool-specific parameters or syntax. Use plain language unless I have given you the tool's syntax. State the format in words. - No generic quality terms such as "masterpiece"; specific description does more. OUTPUT Use the headings Visual decisions, Prompts, Avoid list, If it goes wrong. Put each prompt in its own code block.
What a good response should contain
- A visual decisions list that places the subject on the right with calm, uncluttered space on the left for the headline, with any gaps filled and marked [CHOSEN]
- Prompt A as one paragraph covering the subject, the studio setting, framing, window light, a muted earthy palette and photographic qualities such as shallow depth of field
- Prompt B with a different viewpoint, for example a closer crop on the hands, and Prompt C with the same framing in warmer late-afternoon light
- An avoid list covering text, logos, clutter and heavy saturation
- Troubleshooting notes for likely problems such as distorted hands, a busy left side or an artificial, plastic-looking finish
This is an illustrative scenario showing how to fill in the prompt and what to look for. It is not a recorded model output. Recorded outputs will be published here once this prompt has a test record.
Testing and editorial review
Status: Untested. Reviewed for structure and clarity by the editorial team. No recorded model test yet.
No test record has been published for this prompt yet. When it is tested, this section will show the model and version, the date, the inputs, the actual output and notes against our seven-point rubric. We do not publish numerical scores.
| Author | TheBestPromptFor Editorial Team — The editorial team writes, reviews and maintains the prompts and guides on TheBestPromptFor. Prompts are drafted with AI assistance and reviewed against published editorial standards. |
|---|---|
| Published | November 15, 2026 |
| Last reviewed | November 15, 2026 |
| Last tested | Not yet tested |
| Version history |
|
Limitations and things to check
- This prompt produces text. You need a separate image generator, or a chat tool with one built in, to create the images.
- Image models do not follow descriptions exactly. Expect to generate several images per prompt and to revise the wording.
- Hands, small details, exact counts and any lettering are frequently rendered wrongly and may need retouching or regeneration.
- Rights and permitted uses for generated images depend on the tool's terms and on local law. Check both before commercial use.
- Because the prompt avoids tool-specific syntax, set aspect ratio and similar options in the image tool itself.
Alternative prompts
Graphic Design
Logo Design Brief Prompt
Writes a creative brief and three distinct logo concept directions in words, ready to hand to a designer or adapt for an image tool.
- ChatGPT
- Claude
- Gemini
Graphic Design
Brand Identity System Prompt
Drafts a written identity system: color roles with contrast pairings, a type hierarchy, logo usage rules and imagery direction, tied to your positioning.
- ChatGPT
- Claude
- Gemini
Web Design
Complete Website Design Prompt
A structured brief that walks an AI model from business goals to sitemap, page layouts, visual direction and responsive rules.
- ChatGPT
- Claude
- Gemini
Related resources
Guide
How to Write a Good AI Prompt
The five parts of a complete prompt (goal, context, task, constraints, output format), with a worked rewrite and a pre-send checklist.
Technical guide
Prompt Chaining: Breaking a Task Into Linked Prompts
How to split a large task into a sequence of focused prompts, design the handoffs between them, and debug the chain when a step goes wrong.
Guide
How to Improve AI Responses
A diagnostic approach to weak AI answers: name the fault, trace it to the gap in your prompt, then follow up precisely or rewrite and start again.
Frequently asked questions
Which image generator does this work with?
The prompts are written as plain descriptive paragraphs, which image tools that accept natural-language input can use. The prompt does not add tool-specific parameters unless you paste the syntax you want, because those conventions vary between tools and versions and a text model may recall them incorrectly.
Why not just type a short prompt into the image tool?
You can, and for a quick idea it is fine. A short prompt leaves composition, lighting and color for the image model to decide, so results vary widely. Writing those decisions down gives you images closer to what you pictured and a record of which wording produced them, which matters when you need a matching set.
How do I keep a consistent look across several images?
Run the prompt once, settle on the visual decisions that worked, and paste that list back in as fixed requirements for the next subject. Reusing the same lighting, palette and medium wording word for word is the most dependable way to get images that belong together.
Can I ask for the style of a specific artist?
The prompt is designed not to do that for living artists, studios or copyrighted characters. Describe what you like about the style instead: the medium, the palette, the line quality, the period. That is usually more precise, and it avoids relying on another person's name.
Sources and references
- Prompt engineering guide — OpenAI
- Prompt engineering overview — Anthropic
- Prompt design strategies — Google AI for Developers
Was this prompt useful?
No account needed. Do not include personal information.