A
AiTechWorlds
AiTechWorlds
Midjourney produces some of the highest-quality AI-generated images available. It has a distinctive aesthetic — painterly, detailed, and often breathtaking — and a dedicated community that has developed sophisticated prompting techniques. This guide covers everything from first generation to professional-quality results.
Midjourney is a text-to-image AI: you describe what you want, it generates an image. Unlike Stable Diffusion (which runs locally), Midjourney is a cloud service — all generation happens on their servers.
The current Midjourney experience: Midjourney moved from Discord-only to a web interface (midjourney.com). New users primarily use the web app; Discord is still supported for community and advanced usage.
What it excels at:
What it struggles with:
The core workflow:
A Midjourney prompt has several components:
[Subject] [Environment/Setting] [Artistic Style/Medium] [Lighting] [Mood/Atmosphere] [Technical Parameters]
Weak prompt: a cat
Strong prompt: A sleek black cat sitting on a rain-soaked Tokyo street at night, neon reflections in puddles, cinematic photography, shallow depth of field, atmospheric fog, high contrast
The difference isn't more words — it's more specific instructions on what type of image you want.
Specifying an artistic medium dramatically changes the output:
oil painting — textured, classical painting stylewatercolor illustration — soft, flowing, translucentdigital art — clean, modern illustration stylecinematic photography — film-quality realism, dramatic framingpencil sketch — hand-drawn aestheticconcept art — polished illustration for games/filmsflat design illustration — clean vector-like stylephotorealistic — ultra-realistic photography-style outputLighting has enormous impact on mood:
golden hour lighting — warm, soft, sunset qualitydramatic studio lighting — high contrast, professional photographysoft diffused light — gentle, flattering, no harsh shadowsvolumetric lighting — beams and atmospherebacklit — subject illuminated from behind, silhouette effectneon lighting — cyberpunk, urban, colorful glowcandlelight — warm, intimate, flickering qualitywide angle shot — broad environmental contextclose-up portrait — tight framing on face or subjectaerial view — looking down from aboverule of thirds — compositional balancebokeh background — blurred background, focused subjectDutch angle — tilted camera, tension and uneasemacro photography — extreme close-up detailAdd these to consistently improve output quality:
highly detailed, 8k resolution, award winning photography, masterpiece, trending on artstation, professional photograph
Midjourney parameters go at the end of the prompt with --:
Aspect ratio: --ar 16:9 (landscape), --ar 9:16 (portrait/mobile), --ar 1:1 (square)
Model version: --v 6 (latest Midjourney model), --niji 6 (anime/illustration focused)
Stylize: --s 0 to --s 1000 — how strongly Midjourney applies its aesthetic. Lower = closer to prompt; higher = more artistic interpretation. Default is 100.
Weird: --weird 0 to --weird 3000 — adds unusual, unexpected elements. Good for abstract or creative work.
Quality: --q 1 (standard), --q 2 (more detailed, slower, uses more GPU)
No: --no [element] — exclude specific elements. --no text to remove any text, --no watermark, --no background
Example: A sleek product shot of wireless earbuds on white background --ar 16:9 --s 150 --no shadow
[Product name/description] on [surface/background], studio product photography,
clean white background, professional lighting, commercial photography style,
sharp focus, 8k resolution --ar 3:2
[Abstract concept or visual metaphor], [brand colors if you have them],
flat design illustration, clean modern style, suitable for social media
header image --ar 16:9
[Concept to illustrate], digital illustration, clean editorial style,
pastel color palette, simple composition --ar 16:9
[Scene or concept], concept art, cinematic composition, dramatic lighting,
highly detailed, professional illustration
Professional headshot of a [age] [gender] [profession] person,
business casual attire, neutral studio background, soft professional lighting,
Canon EOS photography --ar 2:3
Note: Avoid generating images of real people without permission.
Getting good images is iterative:
Don't expect the first generation to be perfect. Three to five iterations is typical for professional-quality results.
You can use an existing image as a reference:
--iw (image weight) to control how much the reference influences the output: --iw 0.5 (slight influence) to --iw 2 (strong influence)[image URL] portrait in the same artistic style --iw 1.5
Midjourney Pro subscribers get full commercial rights to their generations. Basic subscribers have more limited commercial rights. Check current terms at midjourney.com — licensing policy has evolved.
Important: Midjourney (and all AI image generators) learn from images including copyrighted work. Don't use AI-generated images as direct substitutes for specific copyrighted artwork, and be thoughtful about style-mimicking specific living artists.
Next lesson: DALL-E 3 and Adobe Firefly — understanding AI image generation for commercial and enterprise use.
Get this course's notes on Telegram!
Free cheat sheets, summaries & practice exercises