Prompt Library

How to Write AI Image Prompts (and 30 to Steal)

30 copy-paste prompts

The 6-part framework that controls every output: Subject, Style, Lighting, Lens, Mood, Parameters. 30 prompts that prove it on Midjourney, DALL-E, and Stable Diffusion.

In short: This page contains 30 copy-paste ready prompts, organized into 4 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Hand-curated and tested by the AI Academy team.

Louis Corneloup
By Louis Corneloup · Founder, Techpresso
Last updated ·Hand-curated & tested by the AI Academy team

The 6-Part Framework (start here)

6 prompts

Subject — the WHAT

1/30

✨ What it does

This prompt is the subject line of an image prompt: you replace the generic slot with a specific person or object so the model has something real to draw. Write that line in detail, then use it as the first piece of your Midjourney or DALL-E prompt.

Subject: [be specific. "A young woman" is generic. "A young woman with auburn hair and freckles, mid-laugh" is generative.] Replace this line with your specific subject.

💡

Pro tip: Specificity is the multiplier. "Person" → generic figure. "40-year-old woman with auburn hair, freckles, mid-laugh" → a specific image. Always add 2-3 specific traits.

Style — the AESTHETIC

2/30

✨ What it does

This prompt is the style slot: you pick or stack an aesthetic such as cinematic photography or Studio Ghibli watercolor. Choose one look, then lock it before you add lighting or camera words.

Style: [Pick one or stack two. Examples: "cinematic photography", "Studio Ghibli watercolor", "1970s film grain", "Art Deco illustration", "by Wes Anderson", "Vogue editorial".] Replace this line with your style.

💡

Pro tip: Named references (named artists, films, magazines, decades) carry more signal than adjectives. "Vogue editorial" beats "high fashion" every time.

Lighting — the MOOD MAKER

3/30

✨ What it does

This prompt is the lighting line that sets mood, from golden hour backlight to a single spotlight. Pick one lighting phrase, then add it so your scary forest can still look peaceful if the light is soft.

Lighting: [Examples: "golden hour backlight", "soft window light from the left", "harsh midday sun with deep shadows", "blue hour twilight", "single spotlight, deep shadows", "softbox lighting from upper-left".] Replace this line with your lighting.

💡

Pro tip: Always specify lighting. The difference between hobbyist AI images and professional ones is usually that the pro versions specify lighting explicitly.

Lens / Camera — the TEXTURE

4/30

✨ What it does

This prompt adds a camera or lens phrase, 85mm, Hasselblad, or Kodak Portra, which shifts the AI's texture even if the physics is fake. Drop one camera line into your prompt, and for DALL-E write it as prose instead of flags.

Lens / camera: [Examples: "85mm f/1.4 lens", "wide-angle 24mm", "macro lens at 3x magnification", "Hasselblad medium format", "shot on Fujifilm X100V", "Kodak Portra 400 film". For DALL-E, use prose: "shot on 85mm lens at f/1.4 for shallow depth of field".] Replace this line with your camera/lens.

💡

Pro tip: "85mm f/1.4" reliably produces shallow depth of field with creamy bokeh — the "professional portrait" look. Memorize this one.

Mood / Color — the EMOTIONAL TONE

5/30

✨ What it does

This prompt is the mood and color line that ties the image together, from warm earth tones to high-contrast monochrome. Choose one palette, then check it does not fight the style you already picked.

Mood / color: [Examples: "warm muted earth tones", "high-contrast monochrome", "pastel color palette", "moody Roger Deakins cinematography", "vibrant primary colors", "faded vintage film tones".] Replace this line with your mood/color.

💡

Pro tip: Pick a mood that matches your subject. A cozy interior with vibrant primary colors will feel wrong — use warm muted tones instead.

Parameters — the CONTROL

6/30

✨ What it does

This prompt lists Midjourney flags for aspect ratio, style raw, niji, version, and stylize, plus prose aspect hints for DALL-E. Add the flags that match the crop you want, then generate and compare raw versus default.

Parameters (Midjourney): --ar 4:5 (portrait) or --ar 16:9 (wide) or --ar 1:1 (square), --style raw (no beauty filter) or --niji 6 (anime), --v 6, --stylize 250 (literal) to 750 (artistic). DALL-E / Stable Diffusion: parameters don't exist — instead, specify aspect ratio in prose ("vertical poster format", "square 1:1 aspect ratio").

💡

Pro tip: Always set --ar (aspect ratio) explicitly. The default 1:1 square is almost never what you want for portraits, landscapes, or wallpapers.

Prompts get you started. Tutorials level you up.

A growing library of 300+ hands-on AI tutorials. New tutorials added every week.

Start 7-Day Free Trial

Subject + Style Recipes

8 prompts

Cinematic Photography Recipe

7/30

✨ What it does

Midjourney turns [Subject] into a cinematic still using [photographer reference], a film color grade, 85mm at f/1.4, and film grain. Fill those slots, then generate at a wide 16:9 crop and keep the frame that feels like your movie still.

[Subject]: [specific person/object/scene], [photographer reference] cinematography, [color grade reference like "Blade Runner 2049 color grade" or "Wes Anderson palette"], 85mm lens, f/1.4, shallow depth of field, film grain, --ar 16:9 --style raw --v 6

💡

Pro tip: Roger Deakins (1917, Blade Runner 2049), Robert Yeoman (Wes Anderson), Hoyte van Hoytema (Dunkirk, Interstellar) are the three most reliable photographer references.

Editorial Magazine Recipe

8/30

✨ What it does

Midjourney builds a Vogue-style editorial of [Subject] with beauty-dish light, a [color] seamless backdrop, and a photographer such as [name like "Tim Walker" or "Mert and Marcus"]. Set the color and photographer, then generate your 4:5 portrait crop.

[Subject], Vogue editorial style, beauty-dish lighting, [color] pastel seamless backdrop, professional makeup, fashion photography by [name like "Tim Walker" or "Mert and Marcus"], --ar 4:5 --style raw --v 6

💡

Pro tip: Beauty-dish lighting is shorthand for "professional but soft." Pair with named fashion photographer for instant editorial energy.

Anime / Ghibli Recipe

9/30

✨ What it does

Midjourney paints [Subject] as Studio Ghibli watercolor in [environment - "lush green hills" or "Tokyo street at dusk"] with a soft hand-painted feel. Pick the setting, then generate with niji and keep the still that looks hand-brushed for your board.

[Subject], Studio Ghibli style watercolor illustration, [environment - "lush green hills" or "Tokyo street at dusk"], soft hand-painted feel, --ar 16:9 --niji 6 --style scenic

💡

Pro tip: `--niji 6` is non-negotiable for anime. Without it, Midjourney defaults to its photoreal model and Ghibli styles come out muddy.

Architectural Photography Recipe

10/30

✨ What it does

Midjourney shoots [Building / interior] like an Iwan Baan architectural photo at [time of day - "dusk" usually best] with a dramatic angle and chosen weather. Set the time of day, then generate wide and pick the frame with the strongest perspective for your set.

[Building / interior], architectural photography by Iwan Baan, [time of day - "dusk" usually best], dramatic perspective, [weather - "clear sky" or "moody overcast"], --ar 16:9 --style raw --v 6

💡

Pro tip: Dusk is the most flattering time for architecture — interior lights are on AND exterior light is still there. Almost never shoot at noon.

Product Photography Recipe

11/30

✨ What it does

Midjourney makes a clean e-commerce shot of [Product] on [surface - "white seamless backdrop" or "polished marble"] with one softbox and a drop shadow. Name the product and surface, then generate square and use the cleanest cutout on your listing.

[Product] on [surface - "white seamless backdrop" or "polished marble"], single softbox light from upper-left, subtle drop shadow, hyper-detailed e-commerce product photography, --ar 1:1 --style raw --v 6

💡

Pro tip: For luxury feel, swap to "polished black marble + directional spotlight + dramatic shadow". For tech, stick with seamless white.

Concept Art Recipe

12/30

✨ What it does

Midjourney renders [Subject - fantasy creature / sci-fi vehicle / environment] as Greg Rutkowski and Ralph McQuarrie concept art with dramatic light and [mood]. Fill subject and mood, then generate wide and save the version you would hang on ArtStation.

[Subject - fantasy creature / sci-fi vehicle / environment], concept art by Greg Rutkowski and Ralph McQuarrie, dramatic lighting, ArtStation trending, ultra-detailed, [mood], --ar 16:9 --v 6

💡

Pro tip: Greg Rutkowski is the single most powerful artist reference in AI image generation. Almost every fantasy concept artist studies his work.

Vintage Poster Recipe

13/30

✨ What it does

Midjourney designs a vintage [decade - "1920s Art Deco" or "1970s sci-fi"] poster of [Subject - travel destination / movie scene / event] with decorative type and a period illustrator. Pick the decade, then generate your 2:3 poster crop for a print.

[Subject - travel destination / movie scene / event], vintage [decade - "1920s Art Deco" or "1970s sci-fi"] poster style, illustrated by [Drew Struzan for movies / decade-appropriate illustrator for travel], decorative typography, --ar 2:3 --v 6

💡

Pro tip: For 1920s use "Art Deco style". For 1950s use "mid-century modern". For 1970s/80s use "Drew Struzan style". Each decade has a defining illustrator.

Watercolor Recipe

14/30

✨ What it does

Midjourney paints [Subject] as a delicate watercolor on white with a soft [color] palette in the manner of a named botanical or children's illustrator. Choose the artist, then generate your 4:5 crop for stationery.

[Subject], delicate watercolor illustration, soft [color] palette, white background, hand-painted feel, by [artist - "Marianne North" for botanicals, "Beatrix Potter" for children, modern names for contemporary], --ar 4:5 --v 6

💡

Pro tip: Watercolor prompts benefit from "hand-painted feel" and "slight paper texture" descriptors. Without them, results read as digital paintings.

Lighting & Mood Recipes

8 prompts

Golden Hour Backlight

15/30

✨ What it does

Midjourney lights [Subject] with golden-hour backlight, a warm rim, sun low behind them, and lens flare in [your style]. Drop in the subject and style, then generate in a 4:5 crop and keep the glowing edge.

[Subject], golden hour backlight, soft warm rim light, sun low in frame behind subject, lens flare, [your style], --ar 4:5 --style raw --v 6

💡

Pro tip: Golden hour = first or last hour of sunlight. The trick: place sun BEHIND the subject (backlight), not in front (front-light).

Window Light — Editorial

16/30

✨ What it does

Midjourney lights [Subject - portrait or interior scene] with soft window light from the [left/right] and gentle shadow falloff in [your style]. Pick the side, then generate and check the falloff still looks like a real window.

[Subject - portrait or interior scene], soft window light from the [left/right], diffused natural light, slight shadow falloff, [your style], --ar 4:5 --style raw --v 6

💡

Pro tip: Specify direction ("from the left"). Without it, light feels flat. Side lighting always beats front lighting for character.

Chiaroscuro — Dramatic Shadow

17/30

✨ What it does

Midjourney lights [Subject] in chiaroscuro with one dramatic source, deep shadows, and a Rembrandt pattern in a painterly oil look. Lock the subject in, then generate and keep the frame where your blacks stay deep.

[Subject], chiaroscuro lighting, single dramatic light source, deep shadows, Rembrandt lighting pattern, painterly oil-painting style, --ar 4:5 --style raw --v 6

💡

Pro tip: Rembrandt lighting specifically refers to the small triangle of light on the shadow-side cheek. Add "subtle triangle highlight on cheek" to lock the look.

Blue Hour — Atmospheric

18/30

✨ What it does

Midjourney sets [Subject - city / interior / landscape] at blue-hour twilight with a purple-blue sky and warm lights just coming on. Name the scene, then generate wide and pick the still you would use for that twenty minutes after sunset.

[Subject - city / interior / landscape], blue hour twilight, soft purple-blue sky, warm interior or street lights starting to glow, contemplative mood, --ar 16:9 --style raw --v 6

💡

Pro tip: Blue hour beats sunset for urban scenes — you get both natural sky color AND warm artificial lights, which sunset lacks.

Studio Three-Point Lighting

19/30

✨ What it does

Midjourney shoots [Subject portrait] with studio three-point lighting, a neutral [color] backdrop, and sharp catchlights. Choose the backdrop color, then generate your 4:5 crop for a headshot.

[Subject portrait], professional studio three-point lighting (key light, fill light, rim light), neutral [color] backdrop, sharp catchlights in eyes, --ar 4:5 --style raw --v 6

💡

Pro tip: Three-point lighting is reliable but predictable. For more interesting portraits, use chiaroscuro or window light instead.

Harsh Midday — Documentary

20/30

✨ What it does

Midjourney photographs [Subject - street scene / candid moment] in harsh midday sun with hard shadows, high contrast, and a Cartier-Bresson documentary feel. Place the subject, then generate a 3:2 crop and keep the frame that uses hard light on purpose in your set.

[Subject - street scene / candid moment], harsh direct midday sunlight, hard shadows, high contrast, documentary photography by Henri Cartier-Bresson, --ar 3:2 --style raw --v 6

💡

Pro tip: Midday light works for documentary/street photography because it forces high contrast. Avoid for portraits — it produces unflattering shadows under eyes.

Overcast Diffused — Editorial

21/30

✨ What it does

Midjourney photographs [Subject - portrait or fashion] in overcast diffused light with no harsh shadows and a muted editorial palette. Put the subject in, then generate and let your styling carry the image.

[Subject - portrait or fashion], overcast diffused natural light, soft even lighting with no harsh shadows, muted color palette, editorial photography mood, --ar 4:5 --style raw --v 6

💡

Pro tip: Photographers call this "nature's softbox." It's flattering on every subject and removes lighting as a complicating variable.

Single Spotlight — Theatrical

22/30

✨ What it does

Midjourney isolates [Subject] in a single overhead spotlight with deep black around them and a theatrical mood. Name who or what stands in the pool of light, then generate and keep the most dramatic isolation for your series.

[Subject], single dramatic spotlight from above, deep black surrounding shadow, theatrical mood, isolated subject in pool of light, --ar 4:5 --style raw --v 6

💡

Pro tip: Spotlight from above creates the strongest drama. Spotlight from the side creates a more cinematic, less theatrical feel.

Like these prompts? There are full tutorials behind them.

Learn the workflows, not just the prompts. 300+ easy-to-follow tutorials inside AI Academy — and growing every week.

Try AI Academy Free

Composition & Camera Recipes

8 prompts

Shallow Depth of Field (Portrait)

23/30

✨ What it does

Midjourney shoots [Subject - portrait] on an 85mm at f/1.4 so the face is sharp and the background goes to creamy bokeh. Fill the subject, then generate a 4:5 crop and keep the one that looks expensive in your set.

[Subject - portrait], 85mm lens, f/1.4 aperture, shallow depth of field, subject sharp with creamy bokeh background, --ar 4:5 --style raw --v 6

💡

Pro tip: 85mm f/1.4 is the gold-standard portrait combination. For more compression (face looks flatter and more flattering), use 135mm f/2.

Ultra-Wide Landscape

24/30

✨ What it does

Midjourney shoots [Subject - landscape scene] on an ultra-wide 24mm with a strong foreground and a sense of vast scale. Add a foreground object, then generate wide and keep the epic frame for your landscape set.

[Subject - landscape scene], ultra-wide 24mm lens, dramatic foreground element, sense of vast scale, deep depth of field, --ar 16:9 --style raw --v 6

💡

Pro tip: Wide-angle lenses make foreground objects feel larger. Always include a foreground element (rock, plant, person) — without it, ultra-wide feels empty.

Telephoto Compression

25/30

✨ What it does

Midjourney compresses [Subject - 2 things at different distances] with a 200mm telephoto so the background feels stacked against the subject. Name both subjects, then generate a 3:2 crop and keep the stacked pair for your edit.

[Subject - 2 things at different distances], shot with telephoto 200mm lens, compressed perspective, background feels close to subject, --ar 3:2 --style raw --v 6

💡

Pro tip: For wildlife, sports, or "mountain looming behind person" shots, telephoto is mandatory. Wide-angle would make the mountain look tiny.

Overhead (Flat Lay)

26/30

✨ What it does

Midjourney shoots an overhead flat-lay of [subjects] on [surface], styled like a magazine spread from straight above. Arrange the items in your prompt, then generate square for a feed or lookbook.

Overhead flat-lay shot of [subjects], top-down 90-degree angle, [surface], styled like a magazine spread, --ar 1:1 --style raw --v 6

💡

Pro tip: Flat lay works for plated food, products, paper goods, and beauty. For drinks, beverages, or layered foods, use "45-degree angle" instead.

Macro — Extreme Close-Up

27/30

✨ What it does

Midjourney makes an extreme macro of [subject] as if on a Canon MP-E 65mm at 3x, with a razor-thin focus plane. Name the tiny subject, then generate a 3:2 crop and keep the sharp sliver for your detail set.

Extreme macro photograph of [subject], Canon MP-E 65mm macro lens at 3x magnification, razor-thin focus plane, --ar 3:2 --style raw --v 6

💡

Pro tip: "MP-E 65mm at 3x magnification" is the specific phrase that triggers macro-photography style. Without "3x" the depth of field is too deep.

Low-Angle Hero Shot

28/30

✨ What it does

Midjourney shoots [Subject - person, building, vehicle] from a low angle looking up so they feel powerful and imposing. Point the camera up at them, then generate a 4:5 crop and keep the hero stance for your campaign.

[Subject - person, building, vehicle], shot from low angle pointing upward, subject feels powerful and imposing, dramatic perspective, --ar 4:5 --style raw --v 6

💡

Pro tip: Low angle exaggerates the subject's power. Use for confident portraits or impressive buildings — avoid for vulnerable or contemplative subjects.

Negative Space — Minimalist

29/30

✨ What it does

Midjourney places [Subject placed in lower-third of frame] under a large empty field on a [solid color] background for a magazine-cover crop. Pick the color, then generate and leave the empty space alone in your layout.

[Subject placed in lower-third of frame], large empty negative space above subject, [solid color] background, minimalist composition, --ar 4:5 --style raw --v 6

💡

Pro tip: Negative space gives editorial sophistication. Specify WHERE the subject sits ("lower-third", "right side") and the empty area appears naturally.

Tracking Shot Motion Blur

30/30

✨ What it does

Midjourney shoots [Moving subject - car / runner / dancer] with a panning blur on the background and a sharp subject in the middle. Name the mover, then generate wide and keep the frame that still reads as speed in your cut.

[Moving subject - car / runner / dancer], shot with panning technique, motion blur on background, sharp subject in middle of frame, dynamic energy, --ar 16:9 --style raw --v 6

💡

Pro tip: For sports/cars: emphasize "motion blur on background, sharp wheels/legs". For dance: emphasize "flowing fabric blur".

Free tool

Image Prompt Builder

Build a rich copy-paste prompt for Midjourney, DALL-E, or Stable Diffusion.

Try it free →

Frequently Asked Questions

The Subject (the WHAT) is mandatory. Everything else is optional — but if you skip Lighting and Style, you get generic results. The 80/20 rule: just adding lighting + style descriptors to a basic subject prompt produces a 10x better image.
Three reasons: (1) AI models are stochastic — same prompt produces different images each run. Generate 4+ variations. (2) Your account settings (style, weirdness, stylize values) affect every prompt. Check `/settings` in Midjourney. (3) Small wording matters more than you'd think. "Soft warm light" vs "soft golden light" produces visibly different outputs. Use exact wording from a prompt that works, then iterate.
Yes — they're the highest-leverage tool in your prompt. Named references compress entire aesthetics into one phrase. "Wes Anderson palette" carries more signal than 20 adjective descriptions. The most reliable references: Roger Deakins (cinematography), Tim Walker (surreal fashion), Greg Rutkowski (fantasy concept art), Wes Anderson (symmetric palette), Drew Struzan (vintage movie posters). Build a reference library as you experiment.
**Midjourney**: comma-separated descriptors + parameters at the end. "Portrait of a woman, golden hour light, 85mm lens, --ar 4:5 --v 6". **DALL-E 3**: full natural-language sentences. "A photograph of a young woman in golden hour light, shot on an 85mm lens with shallow depth of field." DALL-E rewards prose; Midjourney rewards keywords. **Stable Diffusion**: tag-style for anime models, comma descriptors for photo models, always include a negative prompt.
The prompts themselves are free to copy. Whether the IMAGES are commercially usable depends on your generator and subscription tier. Midjourney Pro grants broad commercial rights; Basic is limited. DALL-E 3 grants ownership but OpenAI retains some usage rights. Stable Diffusion is fully open-licensed. Check current terms before publishing.
Yes — every prompt on this page is free to copy and paste. The first prompts are visible immediately with no signup. Drop your email to unlock the rest plus a 10-page guide that goes deeper into each of the 6 framework elements.

Prompts are the starting line. Tutorials are the finish.

A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.

7-day free trial. Cancel anytime.