P
LLOS.ai
Prompt Wizard
AI Art Generator

About & FAQ

About the LLOS.ai Prompt Wizard

The LLOS.ai Prompt Wizard turns simple words into powerful AI image prompts, built step by step. Instead of guessing at syntax, you make a few guided choices — core command, modifiers, boosting, and parameters — and copy a finished, ready-to-paste MidJourney prompt. It is free, runs entirely in your browser, and works on mobile and desktop.

✦ Just have a word? Try the AI Art Prompt Generator →

The five steps

  1. 1Core CommandDescribe the main subject, setting, and action in clear, descriptive language — for example, “a majestic lion on a rocky cliff”.
  2. 2Key ModifiersAdd keywords for style, lighting, and color — oil painting, cinematic lighting, vibrant colors — to define the look and feel.
  3. 3Boosting TechniquesEmphasize or de-emphasize individual concepts with a slider, and list anything to exclude as a negative prompt.
  4. 4ParametersSet technical parameters like aspect ratio (--ar), stylize (--s), and quality (--q) to control shape, creativity, and detail.
  5. 5Mastery & ReviewReview the assembled prompt and copy it to your clipboard, ready to paste into MidJourney.

Frequently asked questions

What is the LLOS.ai Prompt Wizard?
It is a free, browser-based tool that walks you through building one strong MidJourney prompt in five guided steps — Core Command, Key Modifiers, Boosting, Parameters, and Review. Instead of memorising syntax, you make simple choices and the Wizard assembles a finished /imagine command you can copy and paste. For example, starting from “a majestic lion on a rocky cliff”, you add a style and lighting, emphasise the lion, set a 16:9 aspect ratio, and end with a clean, ready prompt. It runs entirely in your browser and works on phone and desktop.
How is it different from the AI Art Prompt Generator?
They are companions for opposite moments. This Wizard is for when you already know your subject and want to refine it into one precise MidJourney command with technical parameters. The AI Art Prompt Generator is for when you have only a word and want dozens of model-agnostic ideas to explore. A typical flow: brainstorm in the Generator (type “grief”, browse twenty directions), pick the one you love, then bring it here to dial in aspect ratio, stylize, and quality. Broad discovery there; exact control here.
How do I write a good MidJourney prompt?
Start with a clear core command — subject, setting, and action in plain words, like “a child playing in a field of flowers”. Then add modifiers for style, lighting, and colour to set the look. Emphasise what matters, exclude what you don’t want, and finish with parameters such as aspect ratio and stylize. The Wizard enforces this order so nothing important is skipped. For example, that child scene becomes stronger as “…oil painting, golden hour lighting, vibrant colours… --ar 3:2”. Specific beats vague: name the medium, the light, and the mood.
What is the Core Command step?
The Core Command is the foundation of your prompt — the essential description of what the image shows: the main subject, the setting, and any action. It should be clear and concrete but still descriptive, for example “a futuristic city skyline at night” rather than just “city”. Everything you add later decorates this core, so a vague core produces a vague image. The Wizard pre-fills a random starter idea to get you going, and you can type your own or edit it freely before moving on. Get this right and the rest builds naturally.
What are Key Modifiers?
Key Modifiers are keywords that define the look and feel of the image, grouped into Style, Lighting, and Colour. They move you from a plain description to a distinctive picture — “oil painting”, “cinematic lighting”, “vibrant colours”. You can select several across the three tabs, and the Wizard appends them to your core command. For example, adding “watercolour”, “soft ambient light”, and “pastel palette” gives a gentle, dreamy result, while “cyberpunk”, “neon lighting”, and “dark and moody colours” gives a hard, glowing one. Each card explains what its modifier does.
What is the difference between style, lighting, and colour modifiers?
They control three independent layers. Style sets the artistic technique or movement (oil painting, anime, art deco). Lighting sets how the scene is lit (golden hour, cinematic, rim lighting), which shapes mood and depth. Colour sets the palette (vibrant, pastel, monochromatic), which drives emotion. Because they are separate, you can mix freely: a “photorealistic” style with “dramatic” lighting and a “sepia” palette feels like an old photograph. The Wizard keeps them in three tabs so you choose deliberately rather than cramming everything into one word, and you can pick several from each.
What are Boosting Techniques?
Boosting lets you control the emphasis of individual ideas — making some concepts more prominent, others less, or excluding them. The Wizard pulls the key words from your core command and gives each a slider. For example, with “a lion and trees on a cliff”, you can push “lion” up and ease “trees” down so the lion dominates. Rather than cryptic numeric weights, the Wizard translates your slider into natural-language emphasis for broad compatibility, so the result stays readable across models. It is how you steer attention without rewriting the whole sentence.
How do I emphasize a concept?
On the Boosting step, each keyword from your prompt has a slider running from de-emphasis to strong emphasis. Drag “dragon” toward the high end and the Wizard rephrases the prompt so the dragon becomes the focal point; drag “smoke” toward the low end to push it into the background. The tool expresses this in plain language — turning a strong setting into a “commanding” or “prime focus” phrase — rather than numeric weights, so it works across different models. Set a neutral value to leave a word untouched. It is a quick way to rebalance a busy scene.
What is a negative prompt (--no)?
A negative prompt tells the model what to leave out. On the Boosting step there is a field where you list things you don’t want — for example people, text, watermarks — and the Wizard adds them as “--no people, text, watermarks” at the end of your command. This is the cleanest way to remove recurring nuisances: if your landscapes keep sprouting tourists, “--no people” clears them. Keep the list short and specific to what actually intrudes; over-stuffing it can confuse the model. It pairs well with emphasis: boost what you want, exclude what you don’t.
What does the aspect ratio (--ar) parameter do?
The aspect ratio, written --ar (or --aspect), sets the shape of the final image. It matters as much as the content: “--ar 16:9” gives a cinematic widescreen, “--ar 9:16” a tall format for phone wallpapers or stories, “--ar 1:1” a square for profile images, and “--ar 3:2” a classic photo crop. For example, a sweeping landscape feels right at 16:9 but cramped at 1:1. The Wizard lets you pick a ratio as a card, then appends it to your command automatically, so you don’t have to remember the exact syntax.
What does stylize (--s) do?
Stylize, written --s (or --stylize), controls how artistically opinionated MidJourney is, on a scale from 0 to 1000. Low values stay literal and faithful to your words; high values let the model add its own flair, colour, and composition. For example, “--s 0” renders close to a plain description, “--s 250” is a balanced default, and “--s 750” becomes lush and interpretive. The Wizard exposes this as a slider so you can feel the trade-off between control and creativity. Reach for low stylize when accuracy matters and high when you want the model to surprise you.
What does quality (--q) do?
Quality, written --q, controls how much rendering effort the model spends — affecting detail and generation time, not the aspect or content. Lower values like “--q 0.25” are fast and rough, good for quickly testing ideas; “--q 1” is the standard; “--q 2” adds refinement for a final image. For example, while exploring twenty compositions you might stay at low quality, then switch to high once you have chosen the keeper. The Wizard offers these as simple cards. Think of it as a draft-versus-final dial: cheap and fast first, polished once you are sure.
Which MidJourney versions are supported?
You can target v7 (latest), v6, or v5.2 using the selector in the header. The version matters because syntax and behaviour change between releases — phrasing that is reliable in one can be inconsistent in another. The Wizard tracks your choice and, when it spots something risky for that version, shows a compatibility heads-up. For example, certain weighted or multi-text constructions behave unevenly in v7, and the Wizard will suggest testing carefully or trying v6 for stability. Pick the version you actually use so the finished prompt matches the model you will paste it into.
Why did I get a compatibility warning?
The Wizard watches your prompt against the MidJourney version you selected and flags constructions that tend to behave unevenly there. For instance, on v7 it warns when your base prompt uses multi-text separators or weighted syntax like “|” or “::”, because those can be inconsistent in that release. The banner is not an error — your prompt still works — it is a heads-up to test carefully or switch to a steadier version such as v6. You can clear it by simplifying the flagged syntax, or simply proceed if you know how that version handles it.
Can I edit my base prompt mid-way?
Yes. On the Boosting step there is an Edit Prompt button that opens your base command in a text box, so you can rewrite the core concept without starting over. This is useful when, halfway through, you realise the subject should change — say from “a lion” to “a snow leopard”. After saving, the Wizard refreshes the keyword sliders to match your new wording, keeping the rest of your choices intact. You can also step Back at any time to revisit earlier steps, or use Restart for a completely fresh prompt. Nothing is locked in until you finish.
What does the final /imagine prompt look like?
It is a single, clean MidJourney command beginning with /imagine prompt:, followed by your boosted description, then your modifiers, any negative prompt, and your parameters. For example: “/imagine prompt: a majestic lion on a rocky cliff, oil painting, golden hour lighting, vibrant colours --no people --ar 16:9 --s 250”. The Wizard assembles all your choices in the correct order so you don’t have to remember where each piece goes. On the final screen you can read it, copy it in one tap, and paste it straight into MidJourney — no editing of syntax required.
How do I copy the finished prompt?
On the completion screen your assembled prompt sits in a text box with a Copy to Clipboard button beneath it. One tap copies the whole command, and the button briefly confirms with a checkmark and a short burst of confetti so you know it worked. From there you paste it directly into MidJourney’s prompt bar. If you would rather make another, the Build Another button resets everything with a fresh starter idea. For example, build a 16:9 hero image, copy it, then immediately build a 9:16 version of the same subject for a phone screen.
Do these prompts work outside MidJourney?
Mostly, with one caveat. The descriptive part — subject, style, lighting, colour — is plain language and transfers well to Stable Diffusion, DALL·E, or Firefly. The parameters (--ar, --s, --q, --no), however, are MidJourney conventions and other models may ignore or rename them. For example, “a watercolour fox, soft ambient light, pastel palette” works anywhere, but “--ar 16:9” may need to be set through that model’s own aspect-ratio control instead. If you want a fully model-agnostic prompt to begin with, use the AI Art Prompt Generator, which writes plain sentences without parameters.
Do I need an account or to pay?
No. The Prompt Wizard is completely free, requires no sign-in, and keeps nothing about you — it runs entirely in your browser. You can build as many prompts as you like, copy them, and start over without any limits or accounts. The only paid part is whatever AI art service you choose to paste the finished prompt into, which is entirely separate from this tool. For example, you can plan a whole shoot’s worth of prompts here for free, then spend your render credits in MidJourney only on the ones you decide to keep.
Can I start over or build another?
Yes, in two ways. Restart (in the footer during the steps) clears your current work and drops you back at the beginning with a fresh starter idea, handy when you want to change direction completely. On the final screen, Build Another does the same after you have copied your prompt, so you can quickly produce a series. For example, finish a cinematic 16:9 landscape, copy it, hit Build Another, and make a square version with a different palette. You can also step Back at any point to adjust a single earlier choice without restarting.
Any tips for stronger results?
A few. Keep your core command concrete — name the subject, setting, and action plainly. Add modifiers deliberately: one strong style, one lighting choice, one palette usually beats a long pile of keywords. Use boosting to make the true subject dominate and a negative prompt to remove recurring clutter. Match the aspect ratio to the scene — 16:9 for landscapes, 9:16 for phone screens. And test stylize: low for accuracy, high for flair. For example, lock your subject and run “--s 100” versus “--s 700” to feel how much creative freedom you actually want.