Quick Answer
Midjourney rewards specificity, so prompts that actually look good follow a clear structure: subject, setting, style, lighting, mood, and parameters. Name a concrete subject, add descriptive style keywords, and tune the technical flags to shape aspect ratio and detail. A vague request gives random output, while a structured one delivers what you pictured.
What Is the Basic Midjourney Prompt Structure?
Midjourney prompts that actually look good come down to specificity, since a structured prompt with the right style keywords and parameters produces exactly what you pictured (arXiv; arXiv). Every good Midjourney prompt follows this order:
[Subject] + [Setting] + [Style] + [Lighting] + [Mood] + [--parameters]
Example: portrait of a young woman, Orchard Road Singapore at night, editorial fashion photography, neon rim lighting, confident and mysterious, --ar 4:5 --v 7 --style raw
Each element does a specific job. Subject and setting tell the AI what to draw. Style and lighting tell it how to render it. Mood guides the emotional quality of the composition. Parameters control the technical output. Skip any element and the AI fills the gap with a random default.
Midjourney only handles images, so if the task in front of you is text instead, like drafting copy or summarising a document, our guide to prompting Gemini for text tasks covers that side of the workflow.
Find all our image generation templates in the PromptCraft Asia image generation category.
Which Midjourney Parameters Are Most Useful?
- --ar [ratio]: Aspect ratio. Use
16:9for landscape,9:16for mobile/portrait,1:1for square,4:5for Instagram. Set this on every prompt. - --v 7: The current model version as of 2026. Always specify to avoid defaulting to an older model.
- --style raw: Reduces Midjourney's default stylization. Produces more realistic, less "AI-looking" output. Essential for product and portrait photography.
- --chaos [0-100]: Controls variation. At 0, all 4 output images are similar. At 100, they are wildly different. Use 20-40 for commercial work, higher when exploring.
- --quality 2: Doubles render time but increases detail. Worth using for final output; unnecessary for quick ideation.
- --seed [NUMBER]: Locks the random seed so regenerations produce similar compositions. Essential for maintaining consistency in a series.
Style Keywords That Work Consistently
These style keywords produce reliable, high-quality output on Midjourney v7:
- photorealistic: Camera-quality output. Pair with
--style rawand a specific camera/lens reference. - cinematic: Wide-angle, cinematic color grading. Think movie still or film poster.
- editorial: Clean, professional magazine quality. Works well for fashion, product, and food.
- minimalist: Clean composition, white or neutral backgrounds, limited elements. Ideal for product photography and logos.
- watercolor: Soft, painted quality. Good for illustrations, character art, and lifestyle content.
- isometric: Flat 3D overhead view. Excellent for icons, diagrams, and tech illustration.
- concept art: Painterly, detailed, fantasy or sci-fi aesthetic. Best for characters and environments.
- product photography: Studio-quality commercial look. Always pair with a specific background color.
Lighting Terms That Transform an Image
Lighting descriptions are among the highest-impact additions to any Midjourney prompt. Use one or two per prompt:
- golden hour: Warm, soft sunset-quality light. Instantly makes outdoor portraits feel premium.
- studio lighting: Clean, even commercial photography lighting with controlled shadows.
- rim light: Light from behind the subject, creating a glowing outline. Dramatic for portraits.
- ambient: Soft, natural environmental light. Subtle and realistic.
- dramatic: High contrast with deep shadows. Strong emotional impact.
- neon: Colorful urban night lighting. Works well for street scenes and tech aesthetics.
- overcast: Diffused, soft, shadow-free. Flattering for skin tones and neutral product shots.
10 Fill-in-the-Blank Midjourney Prompt Templates
Each template follows the full Subject + Setting + Style + Lighting + Mood + Parameters structure. Replace the bracketed variables with your specific content.
- Portrait:
portrait of [SUBJECT_DESCRIPTION], [LOCATION] background, editorial photography, [LIGHTING_TYPE] lighting, [MOOD], shot on Canon EOS R5, 85mm lens, --ar 4:5 --v 7 --style raw - Product photography:
[PRODUCT_NAME] on [SURFACE/BACKGROUND], minimalist product photography, studio lighting, clean shadows, [COLOR_PALETTE] color scheme, commercial quality, --ar 1:1 --v 7 --style raw - Landscape:
[LANDSCAPE_TYPE] in [LOCATION], [TIME_OF_DAY], [WEATHER_CONDITION], [STYLE] photography, [MOOD] atmosphere, wide angle, --ar 16:9 --v 7 - Architecture:
[BUILDING_TYPE] exterior in [CITY/REGION], [ARCHITECTURAL_STYLE], [TIME_OF_DAY] lighting, [SEASON], architectural photography, sharp detail, --ar 3:2 --v 7 --style raw - Social media graphic:
[CONCEPT_DESCRIPTION] for [BRAND_TYPE], flat design, [COLOR_PALETTE], [STYLE: modern/minimalist/bold], white background, social media graphic, no text, --ar 1:1 --v 7 - Logo concept:
minimalist logo concept for [BUSINESS_TYPE], [ICON_DESCRIPTION], [COLOR_1] and [COLOR_2], clean vector style, white background, professional, --ar 1:1 --v 7 - Food photography:
[DISH_NAME] plated [PLATING_STYLE], [TABLE_SURFACE] background, overhead shot, editorial food photography, natural window lighting, [GARNISH_DESCRIPTION], --ar 1:1 --v 7 --style raw - Character:
[CHARACTER_DESCRIPTION], [SETTING_DESCRIPTION], [ART_STYLE: concept art/illustration/3D render], [LIGHTING], [MOOD], detailed, [COLOR_PALETTE], --ar 2:3 --v 7 - Abstract:
abstract [CONCEPT], [COLOR_PALETTE] colors, [STYLE: geometric/fluid/crystalline/organic], [TEXTURE_DESCRIPTION], high resolution, [MOOD], --ar 16:9 --v 7 - Scene:
[SCENE_DESCRIPTION] in [SETTING], [CAMERA_ANGLE: wide shot/close-up/bird's eye], [STYLE] film still, [LIGHTING], [MOOD], cinematic color grading, --ar 16:9 --v 7
What Common Mistakes Should You Avoid?
- Contradicting yourself: "Minimalist but very detailed" confuses the model. Pick one direction.
- Too many style keywords: Stacking 8 style terms dilutes each one. Choose 2-3 that complement each other.
- Forgetting aspect ratio: Default output is square (1:1). Most commercial uses need a different ratio. Set --ar on every prompt.
- Using celebrity names: Midjourney filters famous people. Describe physical characteristics instead.
- No lighting direction: Without lighting terms, output lighting is random. One lighting descriptor doubles the quality of most prompts.
Browse All Image Generation Prompts
The templates above are part of the image generation category on PromptCraft Asia. The full collection includes Midjourney, DALL-E, and Stable Diffusion prompts organized by use case. All prompts have fill-in variable fields and can be copied directly from promptcraftasia.com.