How to Write Text Prompts for AI Image Generators to Get Optimal Results with Ease

ThatPainter is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Learn More

The most effective AI image prompts are not collections of “magic words.” They are clear visual briefs. State what the image is for, identify the subject and action, describe the setting and visual treatment, then add only the composition, lighting, colors, text, and constraints that genuinely matter.

There is no perfect prompt formula that works identically in ChatGPT, Midjourney, Adobe Firefly, Google Imagen, and every other generator. Prompting improves when you treat it as a process: write a focused brief, inspect the result, change one or two variables, and generate again.

What is an AI image prompt?

An AI image prompt is the text instruction used to guide an image generator. It can describe what should appear, what the subjects are doing, where the scene takes place, how the image should look, how it should be framed, and what must remain fixed.

A prompt is guidance, not a deterministic command. The same wording can produce different results, and two image tools may interpret identical wording differently. A strong prompt reduces ambiguity without attempting to describe every pixel.

OpenAI’s image-generation guidance recommends focusing on purpose, subject, action, setting, visual style, framing, lighting, and important constraints. Midjourney’s official prompt guide, by contrast, says that short, simple prompts often work well. Both points are useful: provide the information the image needs, but do not add detail merely to make the prompt longer.

The nine elements of a strong image prompt

Use the following as a flexible checklist, not a rigid formula. Some images need all nine elements; an exploratory landscape may need only three or four.

1. Purpose or use case

Say what the image is for when that affects its layout or visual hierarchy. Purpose helps the generator infer whether you need negative space, a particular orientation, or a polished commercial treatment.

• “A wide website hero image for a sustainability consultancy”
• “An editorial illustration for an article about remote work”
• “A product image for an online store”
• “A square social-media post announcing a summer sale”

2. Main subject

Name the central person, object, animal, place, or concept with concrete nouns and visible attributes.

“A person in a city” leaves too much unresolved. “A young urban cyclist wearing a mustard-yellow rain jacket” gives the model a subject, age range, activity-related identity, and a prominent color.

3. Action, pose, or arrangement

Describe what is happening. Action is especially important for people, animals, sports, and scenes containing several objects.

• “Reading a map while standing beside a vintage motorcycle”
• “Pouring tea into a ceramic cup”
• “Running through shallow ocean water”
• “Looking directly at the camera with a relaxed expression”

Also describe relationships when they matter: “A child holds a red toy while a golden retriever sits beside them” is clearer than “A dog next to a child holding its toy.”

4. Setting and context

Define where the scene occurs and what surrounds the subject: “inside a sunlit Scandinavian kitchen,” “on a misty mountain trail at dawn,” or “at a crowded night market in Taipei.” A setting gives the generator useful environmental and cultural context.

5. Medium or image type

Identify the broad medium or production goal:

• Documentary photograph
• Editorial illustration
• Watercolor painting
• 3D product render
• Film still
• Architectural visualization
• Fashion editorial
• Technical diagram
• Children’s-book illustration
• Ink drawing or collage

Medium is one of the useful prompt categories listed in Midjourney’s official documentation. A medium is more informative than a vague instruction such as “make it professional.”

6. Composition and framing

Include framing when the camera position or layout affects the image’s usefulness:

• Close-up or head-and-shoulders portrait
• Full-body shot
• Wide establishing shot
• Overhead or bird’s-eye view
• Low-angle view
• Centered or symmetrical composition
• Subject on the right side
• Ample empty space on the left for headline text

These terms are interpretive guidance rather than guaranteed camera controls. If the tool offers an aspect-ratio or layout control, use it alongside the wording.

7. Lighting

Describe visible lighting conditions instead of relying on adjectives such as “beautiful” or “amazing.” OpenAI specifically recommends concrete lighting descriptions.

• “Soft natural light entering from a window on the left”
• “Warm sunset backlight with long shadows”
• “A large diffused studio softbox”
• “Cool blue neon light against a dark background”
• “Overcast daylight with low contrast”

8. Color, materials, and texture

Color matters for mood, brand identity, readability, and product presentation. Use a palette such as “muted sage green, cream, and terracotta,” “high-contrast black and electric cyan,” or “warm neutrals with one bright red accent.”

For close-ups and product work, add material and surface information: “brushed aluminum casing,” “rough handmade paper,” “glossy black ceramic,” “velvet upholstery with visible woven texture,” or “dew on translucent flower petals.”

9. Constraints

State what must remain fixed or what the image must accomplish:

• “Leave clean negative space across the upper third.”
• “Show exactly three apples.”
• “Keep the product label facing forward.”
• “No visible brand logos.”
• “Use a horizontal 16:9 composition.”
• “Keep the subject’s clothing colors consistent.”

Important constraints should be stated directly. They still are not guarantees: object counts, lettering, relationships, and precise placement can fail and may need editing.

A simple prompt formula

Use this template as a planning aid:

Create a [purpose/image type] of [main subject] [doing what], in/at [setting]. Use [medium or visual style], with [composition/framing], [lighting], and a [color palette] palette. Include [important details], and leave [required space or constraint].

For example:

Create a wide editorial illustration for an article about sustainable commuting: a young cyclist in a mustard-yellow rain jacket riding through a leafy city street after light rain. Use a clean contemporary magazine-illustration style, a wide side view, soft overcast daylight, muted greens and warm neutrals, and leave clear negative space on the left for a headline.

You do not have to use this exact syntax. A shorter prompt may be better for experimentation or for a tool whose documentation favors concise descriptions.

Before-and-after prompt examples

Product photography

Weak:

A nice picture of a water bottle.

Improved:

A premium product photograph of a matte white stainless-steel water bottle standing on a pale stone surface, soft daylight from the upper left, subtle green leaves in the background, clean modern wellness aesthetic, shallow depth of field, vertical composition, no logos.

Character concept

Weak:

A fantasy warrior.

Improved:

A battle-worn female ranger in layered leather armor standing on a windswept mountain pass, holding a longbow, distant snow-covered peaks behind her, cinematic concept art, cool dawn light, restrained blue-gray and rust palette, full-body three-quarter view.

Editorial illustration

A wide editorial illustration for an article about remote work: two coworkers collaborating at separate desks in a bright apartment, connected by a shared video-call screen, clean contemporary magazine style, soft morning window light, muted blue and warm beige palette, open space on the right for a headline.

Website hero image

A wide website hero photograph for a sustainable architecture studio, a modern timber house surrounded by native grasses on a gentle hillside, viewed from a low three-quarter angle, late-afternoon golden light, natural wood and deep green palette, the house positioned on the right with uncluttered sky on the left for headline text.

Social-media graphic

A square promotional graphic for a summer coffee launch, featuring an iced cappuccino in a clear glass on a bright yellow café table, cheerful editorial photography, strong sunlight and crisp shadows, warm cream and yellow palette, clean empty space above the glass for promotional text, no existing logos.

How long should a prompt be?

Neither “longer is always better” nor “shorter is always better” is true.

Long prompts can introduce contradictory instructions, repeated phrases, irrelevant backstory, and competing priorities. Midjourney’s official guidance explicitly warns that long lists of instructions may confuse the process and says short, simple prompts often work well.

Short prompts are useful when you want the model to make creative decisions. They are inadequate when the image requires a specific layout, exact subject count, brand colors, a defined camera angle, negative space, product orientation, or legible text.

Start with the shortest wording that expresses the goal. Add details only when they resolve a real ambiguity or correct a visible failure.

How to prioritize details

When many requirements compete for attention, organize your thinking in this order:

1. Purpose
2. Main subject
3. Action or arrangement
4. Setting
5. Image type or medium
6. Composition
7. Lighting
8. Color
9. Materials and secondary details
10. Constraints

This is an editorial framework, not a documented universal rule about how every model processes text. It simply keeps the main idea from being buried beneath decorative adjectives.

Use positive descriptions first

Many generators respond more reliably when you describe the desired image instead of listing everything it must not contain.

Rather than writing:

A kitchen with no clutter, no people, no text, no plants, and no dark colors.

Try:

A bright, uncluttered minimalist kitchen with clean white counters, pale oak cabinetry, and a simple architectural interior.

Negative prompting is tool-specific. Midjourney documents a dedicated --no parameter for excluding elements. Natural-language wording such as “no cake” should not be treated as a universal exclusion command. Use a documented control when the tool provides one, and inspect the output.

Prompting people, faces, hands, and bodies

Describe the pose, camera distance, viewing angle, clothing, and visible accessories. “A natural pose” is often more useful than stacking many anatomy-related modifiers.

Common failures include extra fingers, distorted hands, unnatural eye direction, inconsistent identity, merged limbs in group scenes, incorrect clothing details, and ambiguous left/right instructions. Semantically wrong poses can look anatomically plausible, so check the action rather than only the rendering quality.

Start with a simpler composition before adding a crowd or complicated interaction. When identity, pose, or wardrobe consistency matters, use an image reference, character or subject reference where supported, inpainting, or an editing workflow rather than expecting one text prompt to solve everything.

How to generate accurate text inside images

Text is a special case. Specify the exact wording, location, approximate hierarchy, design context, and requirement for legibility:

A clean bakery poster with the exact headline “FRESH EVERY MORNING” in large, dark-green sans-serif lettering across the top, centered above a photograph of fresh bread.

Do not promise perfect spelling or layout. Text rendering has improved, but dense copy, small labels, tables, and complicated infographics remain failure-prone.

Midjourney documents quoted words or phrases for text generation in versions 6 and later, generally with better results for standard Latin characters and English letters. See its text-generation documentation for current behavior.

For professional graphics, the dependable workflow is:

1. Generate the artwork with intentional empty space.
2. Add final copy in Canva, Photoshop, Illustrator, Figma, or another layout tool.
3. Use generated text only for short, prominent wording when you can inspect and correct it.

Aspect ratio, composition, and negative space

Choose the destination before generating. A square image may suit a social post or thumbnail; a portrait crop may suit a story, poster, or book cover; a landscape image may suit a presentation or website hero; and an extra-wide frame may suit a cinematic scene or video thumbnail.

Words such as “wide” and “portrait” help, but use the generator’s aspect-ratio control where available. In Midjourney, --ar is the aspect-ratio parameter. Parameters are separate tool-specific controls, not portable instructions that every image generator understands.

Negative space is often more important than another style adjective. Say where it should be and why: “place the subject on the right and leave clean, low-detail space on the left for a headline.” Then check whether the output still leaves enough usable room after cropping.

How tools differ

ChatGPT image generation

ChatGPT is suited to natural-language instructions and iterative conversation. OpenAI says that one to three clear sentences are often enough, with purpose, subject, action, setting, style, lighting, framing, and constraints added as needed.

During an edit, explicitly name what must not change:

Keep the subject, clothing, color palette, and camera angle unchanged. Move the subject slightly to the right and leave more empty space on the left.

This conversational approach is useful for beginners and for workflows that combine writing, brainstorming, references, and image revision.

Midjourney

Midjourney favors concise descriptive prompts in many situations and provides model-specific parameters, image prompts, style references, subject references, and other controls. Its image-prompt documentation should be consulted for current model behavior and supported values.

Common documented controls include:

• --ar for aspect ratio
• --no for exclusions
• --iw for image-prompt weight, with supported ranges varying by model version
• Quoted words or phrases for text generation in versions 6 and later

Do not assume that Midjourney syntax works in ChatGPT, Firefly, Imagen, or another tool. Model versions and parameter ranges can change.

Adobe Firefly

Firefly is especially relevant when image generation is part of a wider Adobe workflow involving Photoshop, Illustrator, Creative Cloud, editing, or compositing. Adobe’s prompting material frames the process much like briefing a photographer or art director: describe the subject and visual direction, then refine the result.

Adobe also documents prompt-based image generation using GPT Image within Firefly. See the current Firefly help page for interface-specific details.

Google Imagen

Google’s Imagen documentation emphasizes clear descriptions, context, and prompt modification. API users may encounter model-specific parameters and implementation details that do not apply to the Gemini consumer interface, so check the documentation for the exact product and model you are using.

Use reference images when words are inefficient

Text is not always the best way to communicate a precise composition, product shape, clothing design, pose, or color direction. A reference image can carry visual information that would take a long paragraph to describe.

These features are related but not interchangeable:

• Image prompt: influences content, composition, and color.
• Style reference: influences visual treatment.
• Character or subject reference: helps maintain an appearance or identity where supported.
• Editing or inpainting: changes a particular region of an existing image.
• Structure or control reference: helps preserve pose, layout, or depth where supported.

Midjourney describes image prompts as inspiration rather than exact copies and recommends cropping a reference to match the final aspect ratio. Behavior differs substantially across products, models, and versions.

Before uploading a reference, consider copyright, privacy, trademarks, recognizable individuals, and confidential designs. Check the service’s terms, commercial-use rules, privacy policy, and any workplace or client policy that applies.

Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Support on Ko-Fi

A controlled five-pass improvement workflow

Do not rewrite the entire prompt randomly after every disappointing result. Change one or two variables at a time.

Pass 1: Establish the concept

A quiet reading nook beside a large window, warm natural light, editorial interior photograph.

Pass 2: Fix the composition

Keep the reading nook and warm natural light. Use a wide horizontal composition, with the armchair on the right and clear empty space on the left.

Pass 3: Add materials and palette

Keep the composition unchanged. Add a rust-colored linen armchair, pale oak shelving, cream walls, and a small green plant.

Pass 4: Correct one failure

Keep all existing elements and their positions. Remove the extra chair and make the window taller without changing the lighting.

Pass 5: Prepare the asset

Adjust the aspect ratio, resolution, crop, output format, brand colors, object count, background cleanliness, and final text placement. If the model repeatedly fails at one requirement, switch to an editing or compositing workflow instead of adding more adjectives.

Common prompt mistakes

Vague adjectives

“Beautiful,” “epic,” “professional,” and “high quality” are subjective. Replace them with visible choices about framing, light direction, palette, materials, background, and depth of field.

Contradictory instructions

“Minimalist but densely detailed,” “soft diffused light with harsh shadows,” and “centered subject with strong asymmetrical framing” contain competing goals. If the tension is intentional, identify which requirement has priority.

Keyword stuffing

Long lists of camera brands, lenses, artists, styles, and quality terms can obscure the actual goal. Use a few meaningful visual descriptors instead.

Too many subjects

Crowds, group interactions, and scenes with many objects are harder to control than a single-subject composition. Start simple and add complexity gradually.

Relying on negative prompts everywhere

Negative wording is not interpreted consistently. Describe the desired result positively first, then use a documented exclusion control where appropriate.

Changing everything at once

If the subject, style, lighting, layout, and colors all change between generations, you cannot tell which change helped. Keep a short record of each revision and alter one variable when possible.

Assuming prompts transfer perfectly

A Midjourney parameter prompt is not a universal image prompt. Separate portable descriptive content from tool-specific syntax, references, editing commands, and API fields.

When prompting is not enough

Use a different control method when the requirement is more precise than text generation can reliably provide:

• Upload a reference for a product, pose, composition, or color direction.
• Use inpainting or generative fill to correct one region.
• Generate the background and subject separately, then composite them.
• Add final typography in a design application.
• Switch to a tool with stronger editing, layout, or typography features.
• Use a dedicated control or structure reference when pose and geometry are critical.

The best result is often a hybrid workflow rather than a single prompt and a single generation.

Choosing a generator by workflow

There is no universally best image generator. Choose according to the work you need to do.

• Conversational refinement: ChatGPT is a practical choice for beginners, marketers, writers, and users who want to revise images through natural dialogue.
• Artistic exploration: Midjourney suits users who prioritize stylized output, visual experimentation, reference workflows, and specialized controls.
• Adobe production: Firefly is a natural fit for designers and agencies already working in Photoshop, Illustrator, or Creative Cloud.
• Google ecosystem or API work: Gemini and Imagen are relevant to Google users and developers who need Google’s model and tooling ecosystem.

Plans, prices, credits, model names, limits, and interface labels change by product, region, and date. Check the provider’s current pages before subscribing. Useful official starting points include ChatGPT pricing, Midjourney plans, Adobe Creative Cloud pricing, and Google Gemini subscriptions.

AI image prompt checklist

• What is the image for?
• What is the main subject?
• What is happening?
• Where is it?
• What medium or visual treatment should it use?
• How should it be framed?
• What lighting and colors matter?
• What materials or textures matter?
• What must stay fixed?
• What text or exclusions are essential?
• Which aspect-ratio, reference, or editing controls does this tool support?
• What will you change in the next iteration if the result misses the goal?

The practical secret is not a longer prompt. It is a clearer brief, an appropriate tool, and disciplined iteration.

Frequently Asked Questions

How can I keep the same character across multiple images?

Use a character or subject reference, where the generator supports one, and explicitly preserve clothing, colors, age, pose-related features, and other identity details. For difficult sequences, combine references with image editing or inpainting.

Should I name a living artist in an image prompt?

Concrete visual descriptions—medium, palette, lighting, composition, texture, and line quality—are more reproducible and avoid making the result depend on a person’s name. Also consider the generator’s policies and your ethical or legal obligations.

Can I use AI-generated images commercially?

Do not assume commercial safety. Review the generator’s current terms, the rights attached to uploaded references, copyright and trademark issues, likeness rights, jurisdiction, and any client or workplace policy before publishing or selling the asset.

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.