Midjourney v7 Prompt Examples: 50 Tested Prompts (2026)

TL;DR: This guide collects 50 tested Midjourney v7 prompt examples grouped by use case — portraits, products, environments, abstract, illustration, and textures. Each is copy-paste ready with a parameter line and a one-line note on why it works. Adapt any of them by swapping the subject and one or two style modifiers. The fastest path to consistently good output in v7 is a tight formula — subject, two to three style cues, one lighting note, plus parameters — not a longer prompt.

If you have ever stared at the Midjourney prompt bar wondering why your "cinematic, ultra-detailed, 8k, masterpiece" prompt produced something flat and generic, you are not alone. The fix is rarely more adjectives. It is the right structure, the right parameters, and a model-aware understanding of how Midjourney v7 reads your words. These 50 Midjourney v7 prompt examples give you a working library plus the reasoning behind each one, so you can stop guessing and start adapting.

What are the best Midjourney v7 prompt examples?

The best Midjourney v7 prompt examples follow one formula: a clear subject, two to three style modifiers, one or two lighting and composition cues, then parameters. For photoreal images add --raw --s 100; for illustration drop --raw and raise --s to 500-750. Always set --ar. That structure beats long, adjective-stuffed prompts almost every time.

Everything below is built on that formula. Before the prompts, a quick grounding in what changed in v7 — because the model genuinely reads prompts differently than v6 did.

What changed in Midjourney v7?

Midjourney v7 entered alpha on April 4, 2025 and became the platform's default model on June 17, 2025, replacing v6.1 for most users (Midjourney updates). It is not a tweak of the old model — Midjourney describes it as "an entirely new model with unique strengths and probably a few weaknesses" that "may require different styles of prompting" (V7 Alpha announcement). That single sentence is why your v6 prompt library needs a re-test.

Here is what actually shifted, and why it matters for the prompts you write:

• Coherence on hands, bodies and objects improved noticeably. v7 shows "significantly better coherence" in the areas that historically broke AI images (V7 Alpha). You can describe hands and complex poses more directly now.
• Text rendering is stronger. Short labels and signage hold together far better than in v6, though long paragraphs still drift. Keep on-image text to a few words.
• Personalization is on by default. v7 is the first Midjourney model with personalization enabled out of the box, unlockable in roughly five minutes of rating images. Midjourney reports personalization profiles are "preferred by 85% of users" (V7 default model). This means your account's default look may differ from a colleague's on the identical prompt.
• Draft Mode arrived. The --draft flag renders at roughly 10x the speed of a normal generation at about half the cost (V7 Alpha). It is built for rapid iteration before you commit a final render.
• Omni-Reference (--oref) replaced the old workflow for inserting a specific subject into new images, with a dedicated weight control (--ow) (Omni-Reference).

Two practical takeaways. First, because personalization is on by default, two people running the same prompt can get different results — note that when you share prompts with a team. Second, because v7 reads prompts differently, the most reliable approach is a clean, structured prompt rather than a pile of quality-booster words. If you want a deeper primer on writing image prompts from scratch, see our guide to writing AI image prompts.

How do Midjourney v7 parameters work?

Parameters are the dials that sit after your prompt text. Get these right and the same words produce wildly different — and far more controllable — results. Here is the working reference for the parameters used throughout this guide.

Stylize: the single most misunderstood dial

Stylize (--s or --stylize) runs 0 to 1000 and defaults to 100 (Midjourney Stylize docs). Low values keep Midjourney close to your literal prompt; high values "give Midjourney more freedom to interpret your idea," which looks more artistic but can stray from exact details. The mental model: stylize is an artist-freedom slider, not a quality slider. More stylize is not "better" — it is "more Midjourney."

Raw mode: photography's best friend

Raw Mode (--raw) turns off the creative auto-pilot Midjourney normally layers on top of your prompt (Midjourney Raw docs). With simple prompts you get more photo-like results; with detailed prompts you get tighter control over the final look. For anything meant to pass as a real photograph, --raw paired with a low --s is the foundation.

Omni-Reference: the consistency engine

Omni-Reference lets you tell v7 to put a specific thing — a character's face, a product, a recurring object — into a new image. Add --oref [image-url] and tune fidelity with --ow, which ranges 1 to 1000 with a default of 100 (Omni-Reference docs / updates). The practical weight guide:

• 25-100: minimal influence, maximum AI creativity. Use this to restyle a character (photo to anime, for instance).
• 200-400: the popular sweet spot — a balanced blend where faces and clothing stay recognizable.
• 600-1000: strong adherence; objects, faces and style closely match the source. Reserve for when fidelity matters more than creativity.

A useful rule of thumb from the community: unless you are running a very high stylize value, keep omni-weight below 400 or results get unpredictable. We cover full character-consistency workflows in our character consistency guide.

How do you read the prompts in this guide?

Every entry below has three parts: the prompt text, a parameter line, and a short note on what makes it work. The prompts are deliberately written as plain descriptive sentences — that is what v7 reads best — and the parameters do the heavy lifting on look and feel. To adapt any prompt, change the subject first, then swap one or two style modifiers, and only then touch parameters.

A few conventions: lines in code blocks are meant to be copied whole, including the parameter line. Where a prompt would normally include literal on-image text, we mark it (placeholder) because text-in-image is still the least reliable area even in v7 — generate the layout, then add real type in a design tool.

Portrait prompts (10 examples)

Portraits are where v7's coherence gains shine. The improved handling of faces, eyes and hands means you can describe a person directly without over-specifying. Keep stylize moderate and lean on --raw for anything photographic.

1. Editorial portrait

A 30-year-old woman, curly red hair, freckles, wearing a charcoal wool coat, standing in front of a brick wall, soft afternoon light, shallow depth of field, 85mm portrait lens
--ar 4:5 --s 150 --raw --v 7

The 85mm lens cue plus --raw produces natural compression and skin tones. --s 150 keeps it editorial without going painterly.

2. Cinematic close-up

A weathered fisherman, deeply lined face, salt-and-pepper beard, looking off camera, dramatic side light, anamorphic lens flare, 35mm film grain
--ar 21:9 --s 300 --v 7

Dropping --raw here lets Midjourney lean into the cinematic look; the 21:9 ratio sells the widescreen frame.

3. Corporate headshot

Confident female founder, 40s, navy blazer, neutral grey backdrop, soft studio softbox lighting, sharp focus on eyes, clean professional headshot
--ar 1:1 --s 100 --raw --v 7

Low stylize plus --raw keeps it believable for LinkedIn and press. "Sharp focus on eyes" is the single most useful headshot cue.

4. Documentary portrait

Elderly woman tea farmer in Sri Lanka, weathered hands holding tea leaves, golden hour light, environmental portrait, 50mm
--ar 3:2 --s 200 --raw --v 7

v7's hand coherence makes "weathered hands holding tea leaves" actually work now. The 50mm cue gives a natural, non-distorted field of view.

5. Stylized character portrait

Cyberpunk samurai, glowing katana, neon-lit alley behind, holographic tattoos, rim lighting, high contrast
--ar 2:3 --s 600 --c 25 --v 7

High stylize plus chaos 25 generates four genuinely different takes — ideal for character exploration.

6. Soft natural light

Young father holding newborn at window, soft morning light through curtains, intimate moment, shallow depth of field, lifestyle photography
--ar 4:5 --s 150 --raw --v 7

"Lifestyle photography" is a reliable v7 cue for warm, candid framing. --raw keeps it from looking like a stock illustration.

7. Black & white portrait

Jazz musician with saxophone, smoky club background, single overhead light, deep blacks, high contrast, ilford hp5 film grain
--ar 4:5 --s 200 --v 7

Naming a specific film stock (Ilford HP5) reliably nudges grain and tonal response toward classic monochrome.

8. Childhood candid

Boy chasing pigeons in european plaza, motion blur on pigeons, sharp focus on boy, late afternoon golden hour
--ar 3:2 --s 200 --raw --v 7

Pairing "motion blur on pigeons" with "sharp focus on boy" gives Midjourney an explicit depth and motion instruction.

9. Studio fashion

Tall model in oversized red coat, geometric pose, white seamless backdrop, dramatic side rim light, fashion editorial, 85mm
--ar 4:5 --s 250 --v 7

"White seamless backdrop" plus "side rim light" is the bread-and-butter studio setup; v7 handles fabric drape well here.

10. Anime-style portrait

Anime girl with silver hair and turquoise eyes, school uniform, cherry blossom background, makoto shinkai style, soft pastel palette
--ar 4:5 --s 750 --niji 6

Switch to --niji 6 for true anime work. High stylize and a named artist style lock the aesthetic.

Product photography prompts (10 examples)

Product shots reward --raw and controlled lighting language. The job is to make the object look real and desirable, so favor precise lighting cues over decorative adjectives.

11. Hero product on rotating stage

Premium leather wallet centered on dark walnut surface, rotating turntable, dramatic side light, deep shadows, luxury product photography, macro detail
--ar 1:1 --s 150 --raw --v 7

"Macro detail" plus side light pulls out leather grain. Square ratio centers the hero object for catalog use.

12. Lifestyle product

White ceramic coffee mug on linen tablecloth, morning light through kitchen window, blurred toast and book in background, hygge atmosphere
--ar 3:2 --s 200 --raw --v 7

Background props "blurred" tells v7 to keep depth of field shallow and the product in focus.

13. Liquid pour

Whiskey pouring into crystal tumbler, frozen mid-pour, ice cubes, side-lit on dark backdrop, motion frozen, ultra-sharp
--ar 4:5 --s 150 --raw --v 7

"Frozen mid-pour" and "motion frozen" are redundant on purpose — doubling the cue makes the splash crisp.

14. Floating product

Sneaker floating in air, white seamless backdrop, soft drop shadow below, hard rim light from above-right, minimalist commercial photography
--ar 1:1 --s 200 --v 7

Specifying the light direction ("above-right") gives consistent, repeatable shadows across a product line.

15. Top-down flatlay

Watch, leather notebook, fountain pen, espresso cup arranged geometrically on charcoal stone surface, soft north-window light, top-down composition
--ar 1:1 --s 200 --raw --v 7

"Top-down composition" plus "arranged geometrically" produces clean, intentional flatlays rather than random clutter.

16. Tech product hero

Sleek wireless earbuds in matte black charging case, set on brushed concrete, single overhead spotlight, deep matte shadows, premium tech advertising
--ar 16:9 --s 200 --raw --v 7

Matte materials plus a single spotlight read as high-end tech. The 16:9 frame suits web banners.

17. Food editorial

Glistening pomegranate seeds spilling from broken fruit, dark moody background, single side light, water droplets, food photography editorial
--ar 4:5 --s 200 --raw --v 7

"Single side light" on a dark background is the classic moody food look; "glistening" cues the highlight on each seed.

18. Skincare hero

Glass dropper bottle on textured marble surface, eucalyptus leaves, soft natural diffused light, clean minimal composition, skincare brand photography
--ar 4:5 --s 150 --raw --v 7

Low stylize and diffused light produce the calm, clinical-but-warm aesthetic beauty brands use.

19. Fashion accessory close-up

Vintage gold pocket watch on weathered leather, macro detail of engravings, candlelight, warm shadows, antique patina
--ar 1:1 --s 250 --raw --v 7

"Candlelight" and "warm shadows" set the color temperature; "macro detail of engravings" forces fine surface texture.

20. Sports product action

Running shoe mid-stride splashing through puddle, motion blur on water, sharp focus on shoe, dramatic side light, athletic photography
--ar 16:9 --s 250 --v 7

Dropping --raw adds energy and contrast that suits athletic advertising better than flat realism.

Environment and scene prompts (10 examples)

Environments are about scale, atmosphere and a single focal element. Give v7 one clear subject inside the wide shot — a cabin, a figure, a lighthouse — and let lighting and haze carry the mood.

21. Cinematic wide environment

Misty mountain valley at dawn, narrow river winding through, single small wooden cabin, golden first light, atmospheric haze, ultra-wide angle
--ar 21:9 --s 250 --v 7

The "single small cabin" gives the eye a focal point and conveys scale across the 21:9 frame.

22. Architectural interior

Brutalist concrete library, geometric ceiling skylights casting hard light beams, single reader at long oak table, dramatic perspective, vanishing point composition
--ar 16:9 --s 250 --v 7

"Vanishing point composition" is a powerful structural cue v7 honors well, giving strong depth.

23. Tokyo street at night

Narrow Shinjuku alley, neon signs reflecting on wet pavement, salaryman walking away from camera, anamorphic lens flare, blade runner cinematic
--ar 21:9 --s 400 --v 7

Wet pavement reflections double the neon; "walking away from camera" adds narrative without needing a recognizable face.

24. Cozy interior

Warm cabin living room, fireplace crackling, cozy reading nook with stack of books, golden firelight, hygge winter evening
--ar 3:2 --s 200 --raw --v 7

"Golden firelight" and "hygge" lock warmth and softness; --raw keeps it grounded and photographic.

25. Sci-fi cityscape

Vertical megacity at sunset, flying vehicles between superscrapers, layered fog at multiple altitudes, syd mead inspired, retrofuturistic
--ar 21:9 --s 600 --c 20 --v 7

"Layered fog at multiple altitudes" creates atmospheric depth; chaos 20 yields varied city silhouettes.

26. Coastal landscape

Rugged cliffs at golden hour, atlantic waves crashing below, lone lighthouse on point, dramatic stormy sky, long exposure smooth water
--ar 16:9 --s 200 --v 7

"Long exposure smooth water" is a precise photographic instruction that silkens the sea.

27. Forest path

Misty redwood forest at dawn, sunbeams cutting through fog, narrow dirt path leading away, atmospheric depth, soft moss undergrowth
--ar 4:5 --s 250 --raw --v 7

"Sunbeams cutting through fog" (god rays) plus "path leading away" creates depth and an inviting line of sight.

28. Industrial interior

Abandoned factory floor, shafts of dusty light through broken windows, rusted machinery, urban exploration aesthetic, cinematic
--ar 16:9 --s 300 --v 7

"Shafts of dusty light" gives volumetric atmosphere; "urban exploration aesthetic" is a reliable mood shorthand.

29. Desert at night

Sahara desert under milky way, lone tent with warm interior glow, sand dunes in middle distance, long exposure, astrophotography
--ar 16:9 --s 250 --v 7

"Astrophotography" plus "long exposure" cues a star-dense sky; the glowing tent is the warm focal anchor.

30. Underwater scene

Coral reef with school of yellow fish, sunlight rays from surface, clear turquoise water, marine life biodiversity, david doubilet style
--ar 16:9 --s 300 --raw --v 7

Naming a known underwater photographer nudges light quality and color toward documentary realism.

Abstract and mood prompts (5 examples)

Abstract work is where --w (weird) and higher chaos earn their keep. You want texture, color and atmosphere over literal subject matter, so loosen the reins.

31. Liquid texture

Macro of metallic liquid mercury moving in slow motion, geometric reflections, blue and silver palette, abstract texture
--ar 1:1 --s 400 --c 20 --v 7

Defining a tight two-color palette ("blue and silver") keeps abstract output cohesive instead of chaotic.

32. Geometric color study

Risograph print of overlapping circles in coral, mustard, navy on cream paper, slight registration offset, mid-century modern poster
--ar 4:5 --s 500 --v 7

"Slight registration offset" is the detail that makes it read as a real risograph print rather than a digital flat.

33. Ethereal fog

Blue and pink colored smoke swirling on black background, slow motion frozen, dreamlike atmosphere, abstract art photography
--ar 1:1 --s 300 --c 15 --v 7

Black backgrounds make colored smoke pop; modest chaos gives you several distinct swirl patterns to choose from.

34. Particle composition

Golden particles drifting through dark space, lit from behind, cinematic depth, dust motes in light shafts, atmospheric texture
--ar 16:9 --s 300 --v 7

"Lit from behind" creates the glow that makes particles legible against darkness.

35. Surreal landscape

Floating islands above a sea of clouds, twin suns on horizon, scale-dwarfed lone figure on edge, salvador dali meets ghibli
--ar 16:9 --s 600 --c 30 --w 500 --v 7

This is where --w 500 shines — the weird parameter pushes genuinely unexpected, dreamlike combinations.

Illustration prompts (10 examples)

For illustration, drop --raw entirely and raise stylize. Naming a recognizable artist or movement is the strongest single lever you have over style.

36. Children's book illustration

Friendly bear cub in blue raincoat, walking through rainy autumn forest, bright yellow umbrella, watercolor and gouache, beatrix potter inspired, soft palette
--ar 4:5 --s 600 --v 7

"Watercolor and gouache" plus a named illustrator gives a consistent storybook texture across a series.

37. Editorial illustration

Conceptual editorial illustration: figure climbing ladder of books reaching for lightbulb, muted color palette, flat shapes with subtle gradients, christoph niemann style
--ar 1:1 --s 500 --v 7

"Flat shapes with subtle gradients" is the technical cue that produces clean, modern editorial art.

38. Fantasy character art

Elven ranger with longbow, cloak flowing in wind, ancient forest setting, painted illustration, magic the gathering art style, dynamic pose
--ar 3:4 --s 500 --v 7

"Dynamic pose" plus "cloak flowing in wind" injects movement; the named card-art style sets the painterly finish.

39. Comic book panel

Superhero mid-leap between skyscrapers, bold inks, halftone shading, jack kirby retro comic style, dramatic perspective, action lines
--ar 3:2 --s 600 --v 7

"Halftone shading" and "action lines" are the specific tells that make it read as a printed comic page.

40. Art nouveau poster

Woman with flowing hair and flowers intertwining, ornate decorative border, alphonse mucha art nouveau style, gold and emerald palette
--ar 2:3 --s 700 --v 7

High stylize suits ornate movements; the named palette keeps the decorative chaos coherent.

41. Studio ghibli landscape

Lush green hillside with wildflowers, distant village, fluffy white clouds in soft blue sky, studio ghibli aesthetic, hand-painted watercolor feel
--ar 16:9 --s 750 --niji 6

--niji 6 plus very high stylize nails the hand-painted anime-landscape look.

42. Vintage travel poster

Vintage 1950s travel poster, italian coastal village, art deco typography (placeholder), bold flat color palette, screen printed texture
--ar 2:3 --s 600 --v 7

Mark typography as (placeholder) — generate the layout, then set real type in a design tool for crisp text.

43. Botanical scientific illustration

Detailed scientific botanical illustration of lavender, parchment background, latin label below (placeholder), watercolor and ink, victorian naturalist style
--ar 4:5 --s 400 --v 7

Lower stylize here keeps the precise, accurate detail a scientific plate needs.

44. Isometric scene

Isometric tiny village scene, stylized 3d render, pastel color palette, low-poly aesthetic, cute oversaturated diorama, monument valley game inspired
--ar 1:1 --s 600 --v 7

"Isometric" plus "low-poly" produces clean game-art geometry; the named game locks the pastel mood.

45. Pixel art

16-bit pixel art adventurer in mountain pass, retro RPG sprite, limited palette of 32 colors, scanline aesthetic, secret of mana inspired
--ar 16:9 --s 500 --v 7

Stating the bit depth and palette size ("32 colors") keeps the pixel grid honest rather than smoothly rendered.

Background and texture prompts (5 examples)

These are utility assets — surfaces, backdrops and tileable patterns. Use --raw for realism, low stylize, and --tile whenever you need the result to repeat seamlessly.

46. Studio backdrop

Smooth gradient backdrop transitioning from warm orange to deep red, subtle fabric texture, professional studio photography backdrop
--ar 16:9 --s 100 --raw --v 7 --tile

A clean gradient you can drop behind a cut-out product. --tile lets you extend it to any canvas size.

47. Marble texture

Black marble with subtle white veining, polished surface, even soft lighting, seamless tileable texture
--ar 1:1 --s 100 --raw --v 7 --tile

"Even soft lighting" avoids hotspots that would break a repeating tile.

48. Concrete wall

Raw poured concrete wall with subtle imperfections, soft side lighting revealing texture, neutral grey, architectural surface
--ar 16:9 --s 150 --raw --v 7

"Side lighting revealing texture" is the cue that brings out grain instead of producing a flat grey plane.

49. Linen fabric

Natural linen fabric texture, soft folds catching light, neutral oatmeal color, top-down view, organic material texture
--ar 1:1 --s 150 --raw --v 7

"Soft folds catching light" gives dimensional fabric rather than a printed pattern.

50. Watercolor paper

Cream colored cold-press watercolor paper, visible fiber texture, soft side raking light, art supply photography
--ar 1:1 --s 150 --raw --v 7

"Raking light" (light skimming across the surface) is the photographer's trick for revealing paper tooth and fiber.

How do you adapt these prompts to your own subject?

The 50 examples are templates, not endpoints. The reliable adaptation workflow is a four-step swap:

1. Replace the subject. Keep the sentence structure and just change the noun — "premium leather wallet" becomes "ceramic vase." The lighting and parameter scaffolding carries over.
2. Swap one or two style modifiers. Change "luxury product photography" to "rustic farmhouse photography" before touching anything else. Most look changes come from here.
3. Tune one parameter at a time. If realism is off, adjust --s first, then add or remove --raw. Changing two parameters at once makes it hard to learn what did what.
4. Iterate with chaos, then lock with seed. Run --c 20 to see varied directions, pick a winner, then re-run that exact prompt for a tighter set.

A worked example. Start from prompt #11 (the leather wallet) and turn it into a perfume bottle:

Premium glass perfume bottle centered on dark walnut surface, rotating turntable, dramatic side light, deep shadows, luxury product photography, macro detail
--ar 1:1 --s 150 --raw --v 7

You changed one noun and got a usable luxury fragrance shot — same lighting language, same parameters. That is the whole point of a prompt library: the reasoning transfers even when the subject does not.

How do you keep a character consistent across images?

Consistency is the hardest part of any image series, and it is where v7's Omni-Reference changes the game. The workflow:

1. Generate or upload a clean reference image of your character.
2. Add --oref [image-url] to every prompt in the series.
3. Set --ow based on how much you are changing: --ow 300 to keep the face and clothes recognizable across new scenes, --ow 50 if you are restyling (say, turning a photoreal character into an illustration).

Detective in a rain-soaked alley at night, neon reflections, trench coat, cinematic side light
--ar 16:9 --s 200 --oref https://your-image-url.png --ow 300 --v 7

Keep omni-weight under 400 unless you are pushing a very high stylize value, or the model starts producing unpredictable results (Omni-Reference guidance). For a full series-building walkthrough, see our character consistency guide.

How do you use Draft Mode to iterate faster?

Draft Mode (--draft) is the most underused v7 feature for working professionals. It renders at roughly 10x the normal speed at about half the cost (V7 Alpha), which means you can explore ten compositions in the time a single final render used to take.

The workflow that saves the most time:

1. Append --draft to your prompt and fire off variations rapidly — swap subjects, lighting, framing.
2. When a draft composition clicks, re-run the same prompt without --draft for the full-quality final.

Misty mountain valley at dawn, single wooden cabin, golden first light, atmospheric haze, ultra-wide angle
--ar 21:9 --s 250 --draft --v 7

Think of Draft Mode as your sketchpad and standard generation as your final canvas. You would not paint your first idea straight onto the finished canvas — drafts let you fail cheaply and fast.

What are the most common Midjourney v7 mistakes?

After running hundreds of v7 prompts, the same handful of errors account for most disappointing results.

The biggest one is the first. "Cinematic, dramatic, moody, atmospheric, epic, bokeh" does not stack into a better image — the modifiers compete and the model splits the difference into mush. Two or three precise cues beat six vague ones every time. This is the same discipline that makes a good text prompt; if you want the broader principles, our prompt engineering fundamentals post covers them.

What are the power moves for serious Midjourney v7 users?

Once the basics are second nature, these habits separate hobby output from production work:

• Save your top prompts as templates with placeholders. Turn "premium leather wallet" into {{product}} and reuse the scaffold across every product in a catalog. A prompt library beats a clever one-off prompt because it compounds. Tools like Prompt Architects let you save, tag and reuse these with parameter sliders so you are not retyping --ar 4:5 --s 150 --raw --v 7 fifty times a day.
• Pair Omni-Reference with a fixed style. Combine --oref for the character and --sref for the look to lock both subject and aesthetic across a series.
• Mix --raw with low stylize for realism; drop --raw and raise stylize for editorial art. This single trade-off covers most of the realism-versus-style spectrum.
• Use Draft Mode for the messy exploratory phase, then upgrade your winners. Reserve full-quality renders for compositions you have already validated.
• Re-test saved prompts when the default model changes. Midjourney explicitly warns that v7 prompts differently than v6 (V7 Alpha); assume the next version will too.

If your workflow spans more than one model, our cross-model prompt guide explains how to port a look between Midjourney, ChatGPT image generation and others without starting from scratch.

How does Midjourney v7 compare for photoreal versus illustration?

A quick decision table for the two most common goals, so you can pick a parameter starting point without guessing.

The rule underneath the table: stylize and raw are inversely related to realism. The more you want a real photograph, the lower the stylize and the more important --raw becomes. The more you want stylized art, the higher the stylize and the less you need --raw.

Frequently asked questions

What are the best Midjourney v7 prompt examples to start with? Start with a simple subject-plus-modifier formula: subject, two to three style cues, one lighting note, then parameters. For photoreal work use --raw --s 100 and an aspect ratio like --ar 4:5. For illustration, drop --raw and raise --s to 500-750. The 50 copy-paste prompts above are organized by use case so you can grab one and swap the subject.

What's new in Midjourney v7 vs v6? v7 launched in alpha on April 4, 2025 and became the default on June 17, 2025. It ships better coherence for hands, bodies and objects, stronger text rendering, personalization on by default, and a new Draft Mode that runs roughly 10x faster at half the cost. It also added Omni-Reference (--oref) for placing a specific character or object into new images.

How long should a Midjourney v7 prompt be? 20-60 words for most use cases. Below 15 words output gets generic; above 80, conflicting modifiers dilute the signal. The reliable structure is subject, plus two to three style modifiers, plus one to two lighting or composition cues, plus your parameters.

Should I use --raw on every photographic prompt? Yes, whenever photorealism is the goal. Without it, Midjourney applies its house aesthetic, which leans slightly painterly and saturated. Raw Mode turns that off. Pair it with a low stylize value (--s 50-150) for the tightest realism.

What is Omni-Reference and how do I use it? Omni-Reference (--oref [url]) tells v7 to insert a specific character, object or scene element into a new image. Control fidelity with --ow (omni-weight), which ranges 1 to 1000 and defaults to 100. Use 25-100 for stylized reinterpretation and 200-400 for faithful character consistency.

What parameters should I always include in a v7 prompt? Always include --ar. Almost always include --s and --v 7. Add --raw for photographic work, --oref plus --ow for character series, and --draft when you want to iterate fast before a final render.

Can I reuse the same prompt across Midjourney versions? Mostly, but expect different results. v7 produces tighter realism and better coherence than v6 with the same prompt, while v6 leans more saturated and painterly. Midjourney notes v7 may require a different style of prompting, so re-test saved prompts whenever the default model changes.

What does the stylize (--s) value actually change? Stylize controls how much of Midjourney's house aesthetic is applied, on a 0-1000 scale that defaults to 100. Low values stick closely to your literal prompt; high values give the model more artistic freedom and can drift from exact details. Keep it low (50-200) for realism and high (500-750) for illustration and poster art.

By Nafiul Hasan — Founder of Prompt Architects and a daily Midjourney user who has run thousands of v7 generations across product, editorial and illustration work. Last updated: June 10, 2026.