📅 Published June 2025 · 🔄 Updated October 18, 2026 · ⏱ 14 min read
Midjourney prompts changed more with V8 than with any update before it: the default model now rewards full, specific sentences, renders short text, outputs native 2K images, and quietly dropped parameters that most prompt guides still recommend. This guide gives you the prompt formula that works on the current model, 12 hacks from my own Midjourney work, a parameter cheat sheet and 32 copy-paste Midjourney prompts you can run today.
Short answer
Write Midjourney prompts as one clear scene: subject, setting, style, light and camera, then parameters at the end. Put the most important words first, use --raw for realism, and drop --v 6, --cref and --q: V8.2 is the default and doesn't need or support them.
📋 What's inside
- The Midjourney prompt formula for V8
- What changed in Midjourney prompts on V8
- 12 Midjourney prompt hacks that work
- Midjourney parameters cheat sheet
- 32 copy-paste Midjourney prompts
- Consistent characters and products
- Web app or Discord?
- Which Midjourney plan for prompting
- When Midjourney prompts fail
- Using Midjourney prompts responsibly
- If Midjourney isn't the right fit
- FAQ
🔍 How I put this guide together
I've used Midjourney for the image tests on this blog, and the hacks below come from that work. I'm not on a paid plan right now, so I rebuilt every prompt for the current default model on paper rather than rendering each one. Version details, parameter support and prices were checked against Midjourney's official docs on October 18. Where a detail comes from a third-party guide instead, I say so.
Some links are affiliate links: if you buy through them I may earn a commission at no extra cost to you. Midjourney links are not affiliate links. Details
The Midjourney prompt formula that works on V8
- 1
Subject
Who or what, with 2–3 specific details: a matte black ceramic mug, not a mug.
- 2
Setting or action
Where it is and what is happening: on a pale oak table by a window.
- 3
Style or medium
Photo, watercolor, 3D render, poster, line art: name the medium before the mood.
- 4
Light and camera
Light direction, time of day, lens or framing: soft morning light from the left, 85mm, shallow depth of field.
- 5
Parameters
Always last, after a space:
--ar 4:5 --raw. No commas between them.
Here is the same idea written two ways. The first is how most old Midjourney prompts look; the second shows how Midjourney prompts should read on V8.
Before: keyword stack
cat, glasses, bookshelf, cute, 8k, ultra detailed, trending --v 6After: one clear scene
A ginger cat wearing small round glasses sits on the top shelf of a cluttered library bookcase, warm lamp light from below, cozy editorial photo, 50mm lens, shallow depth of field --ar 4:5Tags like 8k, ultra detailed and trending add nothing on V8, because the model already aims for high detail. Spend those words on what makes your image different: the material, the light and the moment.
What changed in Midjourney prompts on V8
| Old habit | On V8.2 | What to do now |
|---|---|---|
--v 6 on every prompt | V8.2 is the default | Leave the version off, or use --v 7 only for V7-only features |
| Comma-stacked keywords | Follows detailed sentences better | Write the scene in full sentences |
--cref / --oref | Not supported on V8.2 | Use the Edit Model with up to four reference images |
--q 2 for detail | Quality parameter not supported | Use --hd for native 2K instead |
| Upscale to get 2K | HD renders 2048px directly | Explore in SD (0.8 GPU min), rerun keepers in HD (1.3 GPU min) |
| Avoid text in images | Short text renders well | Put 1–4 words in double quotes |
| Draft Mode for fast ideas | Draft is V7 only | Switch to V7 for drafts, or explore with chaos in SD |
--niji 6 for anime | Niji 7 is current | Use --niji 7 |
The version history and the support list come from Midjourney's version page. V7 stays selectable, so older Midjourney prompts still run there if you need Omni Reference or quality values.
12 Midjourney prompt hacks that work on V8
--raw for realism, tune --stylize instead of adding adjectives, lock a style with --sref, and iterate with one --seed while changing one thing at a time.1. Front-load the subject
Midjourney weighs the first words most, so strong Midjourney prompts start with the thing you want, then add context.
Red vintage bicycle leaning on a whitewashed wall in a Greek alley, midday sun, travel photo --ar 3:22. Describe light like a photographer
In Midjourney prompts, the direction and quality of light do more than any style word. Say where it comes from and how soft it is.
Portrait of a potter at her wheel, soft window light from the right, dust in the air, warm tones --ar 4:53. Name the lens or framing
Lens and shot type control perspective: 24mm for wide scenes, 85mm for portraits, macro for detail.
Macro shot of a bee on lavender, 100mm macro lens, creamy background blur --ar 1:14. Put short text in double quotes
V8 renders 1 to 4 words reliably. Add where the text sits, like a sign that says.
Hand-painted wooden sign that says "FRESH BREAD" outside a small bakery, morning light --ar 3:25. Exclude in plain English, then use --no
V8 reads exclusions in sentences. Keep --no for stubborn elements.
Empty minimalist beach at dawn with calm water and no footprints --no people, boats --ar 16:96. Add --raw for realism and products
Raw reduces Midjourney's own styling, so your prompt has more say. It was my default for product and portrait shots.
Stainless steel water bottle on a gym bench, gritty natural light, sports product photo --ar 4:5 --raw7. Dial --stylize instead of adding adjectives
Stylize runs 0 to 1000 (default 100). Low values follow the prompt, high values add artistic flair.
Lighthouse on a rocky cliff during a storm, oil painting --ar 2:3 --stylize 6008. Use --chaos to explore, not to finish
Chaos 0 to 100 makes the four images in a grid differ more. Use it early, then drop it to refine.
Logo concept ideas for a mountain coffee brand, flat vector shapes --chaos 409. Lock a look with --sref
Style references copy the look of an image or a style code. --sw sets the strength (default 100).
Cozy bookshop interior at night --sref 2 --sw 150 --ar 16:910. Iterate with one --seed
A fixed seed gives near-identical starting noise, so you can change one word and see its real effect.
Glass perfume bottle on black sand, rim light --seed 4821 --raw --ar 4:511. Explore in SD, finish in HD
SD costs 0.8 GPU minutes per job, HD 1.3. Run ideas in SD and rerun the winner in HD for 2048px output.
Alpine lake at sunrise, mirror reflection, landscape photography --ar 16:9 --hd12. Train your personalization profile
Rate images on midjourney.com to build a profile, then add --p so your Midjourney prompts lean to your taste.
Editorial fashion portrait in a sunflower field, golden hour --p --ar 2:3A note on hack 9: --sref 2 is a short numeric style code; you can paste an image URL or any code you find instead. Codes behave differently across versions, so test one on two or three Midjourney prompts before using it for a whole series.
Midjourney parameters cheat sheet (copy-ready)
--ar, --raw, --stylize, --chaos, --no, --seed, --sref, --p and --hd. Tap Copy and paste any of them onto your prompt.⛔ Not on V8.2
--oref, --ow, --cref, --cw, --q, Draft Mode and multi-prompts with ::.
🔁 Use instead
The Edit Model for references, --hd for detail, and full sentences for weighting.
Want Midjourney prompts with the parameters picked for you? My free AI image prompt generator builds a full prompt from a few choices, with a Midjourney format option, so you can paste it straight in.
32 copy-paste Midjourney prompts by use case
📦 Product
A matte black ceramic coffee mug on a pale oak table, soft morning window light from the left, shallow depth of field, clean product photography with a light gray backdrop --ar 4:5 --rawAmber glass skincare dropper bottle standing on wet river stones, water droplets on the glass, soft diffused studio light, premium beauty ad --ar 1:1 --rawWhite wireless earbuds case floating above a pastel blue surface, soft shadow below, top-down studio shot, e-commerce hero image with empty space on the right --ar 16:9 --rawRunning shoe in mid-air with bursts of colored powder, high-speed flash photography, dark background, dynamic sports ad --ar 3:2 --stylize 250👤 Portrait
Close-up portrait of a woman in her 40s with silver-streaked hair, soft window light, natural skin texture, calm expression, 85mm lens, shallow depth of field --ar 4:5 --rawProfessional headshot of a smiling man in a navy blazer, light gray studio background, soft key light, business profile photo --ar 1:1 --rawElderly fisherman mending a net in a small harbor at golden hour, weathered hands in sharp focus, documentary photography, warm tones --ar 3:2Young chef in a busy restaurant kitchen, steam and motion blur behind her, confident look at the camera, cinematic lighting --ar 16:9📱 Social & marketing
Flat lay of a laptop, notebook, iced coffee and a small succulent on a white desk, bright airy light, lifestyle photo with empty space for text at the top --ar 4:5Bold thumbnail background of a surprised golden retriever next to a giant stack of books, saturated colors, clean empty space on the left --ar 16:9 --stylize 300Cozy reading nook with a window seat, rain on the glass, warm lamp light, soft neutral palette, vertical composition for a pin --ar 2:3Minimal promotional banner of abstract flowing ribbons in coral and teal on an off-white background, lots of negative space --ar 3:1🏛️ Interior & architecture
Scandinavian living room with a light gray sofa, oak floors, tall windows, afternoon sun and indoor plants, interior design magazine photo --ar 16:9 --rawModern cabin with floor-to-ceiling glass in a snowy pine forest at blue hour, warm lights glowing inside, architectural photography --ar 3:2Small home office in a converted attic with a sloped ceiling and skylight, built-in shelves, warm minimal style, wide-angle interior shot --ar 4:3 --rawConcrete museum courtyard with a still reflecting pool, overcast soft light, symmetrical composition, architecture photography --ar 16:9🎨 Illustration & art
Watercolor illustration of a fox reading a book under a mushroom in the rain, loose brushstrokes, soft pastel palette, picture book style --ar 4:3Isometric 3D illustration of a tiny floating island with a lighthouse and a café, soft clay render, bright cheerful colors --ar 1:1 --stylize 400Retro travel poster of a coastal Italian village at sunset, bold flat shapes, limited palette of orange, teal and cream, screen-print texture --ar 2:3Anime-style girl riding a bicycle along a rice field at dusk, fireflies around her, gentle wind in her hair --niji 7 --ar 16:9🍞 Food & drink
Rustic sourdough loaf on a linen cloth with a light flour dusting, warm side light, artisan bakery mood, overhead food styling shot --ar 4:5Overhead shot of a colorful poke bowl with salmon, avocado and edamame on a dark slate surface, natural light, food blog photography --ar 4:5 --rawEspresso pouring into a glass cup, crema swirling, backlit steam, dark moody café background, macro shot --ar 3:2Stack of fluffy pancakes with blueberries and maple syrup dripping, bright breakfast table, morning light, shallow depth of field --ar 4:5🔤 Text & logos
Vintage enamel sign on a brick wall that says "OPEN LATE", warm neon glow, night street photography --ar 3:2Minimal flat logo mark of a paper plane inside a circle, single navy color on white, vector style, no gradients --ar 1:1 --rawPaper coffee cup sleeve printed with the words "MORNING FUEL" in bold hand-lettered type, product mockup, soft studio light --ar 4:5Book cover with the title "THE QUIET HOUR" in elegant serif letters over a misty mountain lake at dawn, centered composition --ar 2:3🧩 Backgrounds & patterns
Repeating tile pattern of lemons and olive leaves on a cream background, Mediterranean tile style --tile --ar 1:1Soft gradient mesh background in lavender and peach with subtle grain, calm and modern website hero --ar 16:9Abstract texture of dark green marble with thin gold veins, high detail, presentation slide background --ar 16:9Wallpaper of a misty mountain range at sunrise, layered silhouettes, soft blue and pink sky, minimal --ar 16:9 --stylize 200To adapt any of these Midjourney prompts, change only the subject first and keep the light, camera and parameters. If the result drifts, move your subject words closer to the start before you touch anything else.
How to keep characters and products consistent
--sref for a fixed style, a saved character description you reuse word for word, and one --seed per series.🖼️ Edit Model
Upload up to four references on the web; on Discord, use --edit with image URLs.
🎨 Style reference
One --sref code across a series keeps colors and rendering the same.
📝 Fixed description
Keep a 20–30 word character sheet: face, hair, outfit, age. Paste it unchanged into your Midjourney prompts.
🔢 One seed
Reuse a seed while you change pose or setting, so the rest stays steady.
For the full workflow, with examples of what breaks and how to fix it, see my Midjourney character consistency guide.
Midjourney web app or Discord?
| Web app | Discord | |
|---|---|---|
| Best for | Most people, visual editing | Power users who like commands |
| Prompt bar and gallery | Yes | Bot commands |
| Editor and Edit Model | Visual tools | --edit with image URLs |
| Conversational Mode | Yes | No |
| Video | Yes | --video |
Which Midjourney plan do you need for prompting?
| Plan | Monthly | Yearly (per month) | Fast GPU time | Relax Mode | Stealth |
|---|---|---|---|---|---|
| Basic | $10 | $8 | 3.3 hr | No | No |
| Standard | $30 | $24 | 15 hr | Unlimited images | No |
| Pro | $60 | $48 | 30 hr | Images + SD video | Yes |
| Mega | $120 | $96 | 60 hr | Images + SD video | Yes |
Prices are from Midjourney's plan comparison, in USD. One rule to know before client work: a company with more than $1 million in gross yearly revenue must be on Pro or Mega.
Since HD costs 1.3 GPU minutes against 0.8 for SD, testing new Midjourney prompts in SD stretches Fast hours by roughly a third more jobs. That habit matters most on Basic, which has no Relax Mode to fall back on.
Midjourney Standard · 15 Fast hours plus unlimited Relax images; $24 a month on yearly billing.
See Midjourney plans →Paying for several AI tools already? My AI subscription cost calculator totals them in a minute, and my guide to the AI subscriptions worth paying for compares Midjourney with the others.
When Midjourney prompts fail: 5 quick fixes
| Symptom | Likely cause | Fix |
|---|---|---|
| Subject is wrong or missing | It sits at the end of the prompt | Move it to the first 3–5 words |
| Muddy, mixed look | Three or more competing styles | Keep one main style and one accent |
| Parameter does nothing | --cref, --q or :: on V8.2 | Use the Edit Model, --hd or sentences |
| Garbled lettering | Too many words or single quotes | 1–4 words in double quotes; try --raw |
| Four random images | Chaos too high | Lower --chaos or remove it |
For deeper troubleshooting, my guide to fixing failing Midjourney prompts covers seven fixes in detail, and how to make AI images that don't look fake tackles the plastic, over-smooth look.
Using Midjourney prompts responsibly
Describing a look works better than naming a living artist anyway: loose watercolor, soft pastel palette, picture book style tells the model more than a name does. That's why none of the Midjourney prompts in this guide name an artist, a studio or a film.
🎨 Describe, don't copy
Name the medium, palette and mood instead of a living artist or brand.
🏷️ Disclose AI
Say when work is AI-made, especially for clients and products you sell.
🧑 No real people
Don't create misleading images of real people or fake news scenes.
⚖️ Know the limits
In the United States, the Copyright Office won't register purely AI-generated images, so add your own edits.
If privacy is part of your concern, my AI privacy settings guide shows what to switch off in the main AI apps.
If Midjourney isn't the right fit
Midjourney has no free plan, so if you want to practise Midjourney prompts before paying, these two take the same prompt structure and offer free credits to test with.
OpenArt
Many image models plus style and character tools
OpenArt hosts several popular image models side by side, so you can paste one of the prompts above and compare outputs. It suits people who want to test models before committing to one.
ImagineArt
Daily free credits · web, iOS and Android
ImagineArt gives a daily pool of free credits across a different set of models. Free images can go to a public gallery, so keep client work on a paid plan.
More options: my best free AI image generators list, Midjourney vs DALL-E, Leonardo vs Midjourney and ChatGPT vs Gemini vs Seedream compare them head to head.
My take on writing Midjourney prompts now
The biggest upgrade most people can make to their Midjourney prompts is to stop writing for V6. Drop the version flag and the dead parameters, describe one clear scene in full sentences, and let --raw, --stylize and --sref do the fine-tuning.
Start with the library above, change one subject at a time, and save the Midjourney prompts that work as your own templates. That habit beats any single hack.