AI Image Prompts: Complete Guide to Midjourney, DALL-E, and Stable Diffusion

AI image generation has transformed creative work—from concept art to marketing visuals to personal projects. But the difference between mediocre and stunning results often comes down to how you write your prompts.

This comprehensive guide covers prompt engineering for all major image generators, with platform-specific tips and hundreds of technique examples.

Understanding Image Prompt Structure

The Basic Formula

Every effective image prompt follows this core structure:

[Subject] + [Description] + [Style] + [Technical Parameters]

Example:

A wise elderly wizard, long silver beard, holding a glowing crystal staff,
in a mystical forest at twilight, 
digital art style inspired by Greg Rutkowski,
dramatic lighting, highly detailed, 8k

The Expanded Formula

For more control, use this comprehensive structure:

[Medium] [Subject] [Subject details] [Setting/Background] 
[Composition] [Lighting] [Mood] [Style] [Artist influence] 
[Technical quality] [Platform parameters]

Let's break down each element:

Element	Purpose	Examples
Medium	Physical form	Oil painting, photograph, 3D render, watercolor
Subject	Main focus	A woman, a dragon, a coffee shop
Subject details	Specifics	Red dress, blue scales, vintage aesthetic
Setting	Environment	In a forest, on a spaceship, during sunset
Composition	Framing	Close-up portrait, wide shot, centered
Lighting	Light quality	Dramatic, soft, golden hour, neon
Mood	Emotional tone	Mysterious, joyful, melancholic
Style	Artistic direction	Photorealistic, anime, minimalist
Artist influence	Reference styles	In style of Monet, Artgerm, Wes Anderson
Technical quality	Resolution/detail	8k, highly detailed, sharp focus
Parameters	Platform-specific	--ar 16:9, --v 6 (Midjourney)

Midjourney Prompt Techniques

Core Midjourney Structure

/imagine prompt: [Your prompt here] --[parameters]

Essential Parameters

Parameter	Function	Example
--ar	Aspect ratio	--ar 16:9, --ar 9:16, --ar 3:2
--v	Version	--v 6, --v 5.2
--style	Style intensity	--style raw (less MJ aesthetic)
--q	Quality	--q 2 (more detail, slower)
--s	Stylize	--s 100 (default) to --s 1000 (very stylized)
--c	Chaos	--c 0 to --c 100 (more variation)
--no	Negative prompt	--no text, watermark, blurry

Midjourney Prompt Examples

Cinematic Portrait:

/imagine prompt: Close-up portrait of a cyberpunk detective, 
neon lights reflecting in wet streets behind her, 
rain droplets on leather jacket, 
cinematic lighting, anamorphic lens, 
blade runner aesthetic --ar 2.39:1 --v 6 --s 200

Product Photography:

/imagine prompt: Luxury perfume bottle on marble surface, 
soft morning light through sheer curtain, 
elegant minimalist composition, 
studio photography, 
commercial quality --ar 4:5 --v 6 --style raw

Fantasy Landscape:

/imagine prompt: Ancient elven city built into crystalline cliffs, 
bioluminescent waterfalls, 
multiple moons in the sky, 
epic scale, 
matte painting style, 
detailed fantasy art --ar 16:9 --v 6 --s 350

Character Design:

/imagine prompt: Character design sheet, 
steampunk inventor, young woman, 
clockwork arm prosthetic, goggles on forehead,
multiple views: front, side, back,
white background, 
concept art for video game --ar 3:2 --v 6

Midjourney Style Keywords

Photography Styles:

• photography, photorealistic
• DSLR photo, portrait photography
• macro photography, long exposure
• Hasselblad, Leica, Sony camera

Artistic Styles:

• oil painting, watercolor, acrylic
• digital art, concept art
• anime, manga, comic book
• art deco, art nouveau, baroque

Lighting Styles:

• golden hour, blue hour
• dramatic lighting, rim lighting
• soft diffused light, hard shadows
• neon lighting, studio lighting

DALL-E 3 Prompt Techniques

DALL-E 3 Strengths

DALL-E 3 excels at:

• Following complex instructions precisely
• Rendering accurate text in images
• Photorealistic renders
• Maintaining consistency with detailed descriptions

DALL-E 3 Best Practices

1. Be Conversationally Detailed

DALL-E 3 understands natural language better than keyword lists:

Create a photorealistic image of a cozy home office in the early morning. 
The desk is positioned by a large window with soft daylight streaming in. 
There's a steaming cup of coffee in a ceramic mug, a MacBook Pro, 
some potted plants including a small monstera, and a stack of books. 
The aesthetic should be minimalist Scandinavian with warm wood tones.

2. Specify Text Explicitly

DALL-E 3 can render text accurately:

A vintage movie poster for a film called "THE LAST HORIZON" 
featuring a lone astronaut standing on a cliff overlooking 
a massive alien landscape. The title text should be in 
bold retro typography at the top, with "COMING SUMMER 2026" 
at the bottom. 70s sci-fi aesthetic.

3. Request Specific Formats

Create a square Instagram-style food photo of a gourmet 
cheeseburger with dramatic lighting, shallow depth of field, 
and the burger positioned slightly off-center. 
Professional food photography style.

DALL-E 3 Prompt Examples

Logo Design:

A minimalist logo for a tech startup called "APEX" - 
the design should incorporate a subtle mountain peak shape 
formed from negative space. Use clean lines, 
modern geometric shapes. The text APEX should be 
integrated into the design in a bold sans-serif font. 
White background, using only royal purple and white colors.

Book Cover:

Design a book cover for a psychological thriller titled 
"THE MIRROR'S EDGE" by Sarah Chen. The design should 
feature a shattered mirror with someone's eye visible 
through a crack, creating an unsettling atmosphere. 
Dark, moody color palette with pops of red. 
Title in elegant serif typography at the top.

Social Media Graphics:

Create a LinkedIn carousel thumbnail for a post about 
"5 Leadership Mistakes to Avoid" - clean corporate design 
with the number 5 prominently displayed, professional blue 
color scheme, subtle geometric patterns in background. 
Modern, credible business aesthetic.

Stable Diffusion Prompt Techniques

Stable Diffusion Syntax

Stable Diffusion uses a different approach with:

• Positive prompt (what you want)
• Negative prompt (what to avoid)
• Emphasis weighting with parentheses: (important word:1.3)
• Models (checkpoints) that change style

Positive and Negative Prompts

Positive Prompt:

(masterpiece, best quality:1.4), 1girl, solo, 
long flowing silver hair, intricate blue eyes,
elegant ballgown, fantasy castle ballroom background,
crystal chandeliers, magical atmosphere,
highly detailed face, 8k, sharp focus

Negative Prompt:

(worst quality, low quality:1.4), blurry, ugly, 
deformed, distorted, extra limbs, missing limbs,
bad anatomy, bad hands, bad fingers,
watermark, text, signature, logo

Weight Syntax

(keyword)       - slight emphasis
(keyword:1.2)   - 20% more emphasis  
(keyword:1.5)   - 50% more emphasis
(keyword:0.8)   - 20% less emphasis
[keyword]       - slight de-emphasis

Stable Diffusion Examples

Anime Character:

Positive:
(masterpiece, best quality:1.4), anime girl, 
pink twin tails, large expressive eyes, 
school uniform, sakura blossoms falling,
soft lighting, (detailed face:1.2),
studio ghibli style, warm colors

Negative:
(worst quality:1.4), (low quality:1.4), 
simple background, bad anatomy, deformed,
extra fingers, watermark, cropped

Photorealistic Portrait:

Positive:
RAW photo, professional photography,
beautiful woman, 25 years old,
looking at camera, soft smile,
natural makeup, (detailed skin texture:1.1),
(bokeh background:1.2), golden hour lighting,
8k uhd, dslr, high quality

Negative:
(painting, illustration:1.3), fake, 
oversaturated, overexposed, blur,
bad anatomy, plastic skin, 
watermark, text, logo

Style Reference Library

Photography Categories

Portrait Photography:

portrait photo, DSLR, 85mm lens, shallow depth of field,
natural light, soft shadows, crisp focus on eyes,
rembrandt lighting, beauty dish

Product Photography:

product photo, studio lighting, white seamless background,
softbox, reflector, commercial quality, sharp focus,
color-accurate, professional product shot

Street Photography:

street photography, candid, urban environment,
35mm film aesthetic, natural lighting, 
decisive moment, documentary style

Artistic Movements

Impressionism:

impressionist painting, visible brushstrokes,
light and color focus, natural scenery,
Monet style, soft edges, atmospheric,
dappled light, en plein air

Art Nouveau:

art nouveau illustration, organic flowing lines,
decorative borders, nature motifs, 
Alphonse Mucha style, ornate details,
muted color palette, elegant curves

Cyberpunk:

cyberpunk aesthetic, neon lights, rain-slicked streets,
holographic advertisements, high-tech low-life,
blade runner style, synthwave colors,
chrome and metal, urban dystopia

Fantasy Art:

epic fantasy art, magical atmosphere,
dramatic lighting, intricate details,
concept art, digital painting,
lord of the rings style, mythical creatures

Cultural Aesthetics

Japanese/Anime:

anime style, cel shading, vibrant colors,
manga influence, expressive eyes,
japanese illustration, light novel art

Vintage/Retro:

vintage aesthetic, retro color palette,
film grain, nostalgic, mid-century modern,
1960s advertising style, aged paper texture

Minimalist:

minimalist design, negative space,
clean lines, limited color palette,
simple composition, elegant, 
scandinavian aesthetic

Common Mistakes and Fixes

❌ Too Vague

Bad: "A pretty landscape"

Good: "A sweeping Icelandic valley at sunset, dramatic cloud formations reflecting golden and pink light, a winding river cutting through volcanic black sand, epic scale landscape photography, wide angle lens"

❌ Conflicting Instructions

Bad: "A realistic cartoon of a person"

Good: "A stylized character portrait with realistic proportions but illustrated in clean vector art style with bold outlines"

❌ Kitchen Sink Prompting

Bad: "A warrior princess in space with dragons and also it's underwater and there are zombies and make it look like van gogh but also like a photograph"

Good: Focus on one clear concept, then iterate

❌ Ignoring Platform Strengths

Bad: Using keyword-heavy Stable Diffusion syntax for DALL-E 3

Good: Match your prompt style to the platform:

• DALL-E 3: Natural language descriptions
• Midjourney: Artistic keywords + parameters
• Stable Diffusion: Weighted keywords + negative prompts

Advanced Techniques

Image-to-Image Prompting

When using an image as reference:

[Your text prompt], inspired by uploaded image,
maintain the color palette, similar composition,
but change [specific elements you want different]

Consistent Character Design

For characters across multiple images:

[Character name], [detailed appearance description],
[consistent outfit details], [distinctive features],
in [specific scene/setting], maintaining character sheet reference

Upscaling Prompts

When generating with upscaling in mind:

[Your prompt], highly detailed, sharp focus,
intricate details, 8k resolution, 
crisp edges, no blur, no artifacts

Prompt Templates by Use Case

Marketing Materials

Professional [type of material] for [brand/product],
[brand colors] color scheme, modern clean design,
[target audience] appeal, [mood/feeling],
commercial quality, high-end aesthetic,
white background, centered composition

Social Media Content

Eye-catching [platform] post about [topic],
[aspect ratio for platform], 
trending aesthetic, shareable,
[brand style], [emotion to evoke],
optimized for engagement

Concept Art

Concept art for [project type], [subject],
[setting/world details], [mood],
multiple views/variations,
industry standard quality,
ready for production reference

Frequently Asked Questions

Which AI image generator is best?

• Midjourney: Best overall artistic quality and aesthetics
• DALL-E 3: Best for following precise instructions and text rendering
• Stable Diffusion: Best for customization, free/local use, and specific model control

How do I get consistent characters?

Unfortunately, pure prompt-based consistency is challenging. Options:

• Use reference images (img2img)
• Use consistent detailed descriptions
• Use character-focused models (Stable Diffusion)
• Use Midjourney's --sref style reference

Can I use AI images commercially?

• Midjourney: Yes, with paid plans
• DALL-E 3: Yes, with usage rights per OpenAI terms
• Stable Diffusion: Yes, depending on model license Always check current terms for your specific use case.

How do I avoid "AI look"?

• Use --style raw in Midjourney
• Add imperfection keywords: film grain, natural lighting
• Reference specific photography techniques
• Avoid "ultra realistic, 8k, perfect" overuse

Conclusion

Mastering AI image prompts is an iterative skill. Start with clear, detailed descriptions, learn platform-specific syntax, and continuously refine based on results.

Key takeaways:

1. Structure matters: Use the formula: Subject + Description + Style + Parameters
2. Platform-specific: Match your prompt style to the AI generator
3. Details win: Specific beats vague every time
4. Iterate: Generate, analyze, refine, repeat
5. Learn from artists: Study artistic terminology and styles

---

Want to explore more creative AI techniques? Browse our creative prompts collection or read our complete prompt engineering guide.