Midjourney V7 Complete Guide: From First Image to Professional Output

Midjourney V7 represents a significant leap in AI image generation — not just in output quality, but in controllability. Photographers, illustrators, and brand designers who previously dismissed AI tools are adopting V7 for production work. Here is how to use it effectively.

Understanding How Midjourney Reads Prompts #

Midjourney reads prompts left to right, with earlier tokens weighted more heavily than later ones. Subject, then style, then technical parameters is the structure that produces the most predictable results. "A weathered lighthouse at sunset, oil painting style, dramatic sky, warm colors, cinematic composition" — subject first, then modifiers in order of importance.

The V7 Prompt Formula #

Consistent structure produces consistent quality: Subject, Action or State, Environment, Lighting, Style, Mood or Atmosphere. Not every element is required, but filling in more dimensions gives the model more to work with. "An architect reviewing blueprints at a drafting table, late evening office, warm desk lamp light, architectural sketch style, focused and contemplative."

Style References (--sref)

The V7 Style Reference parameter is a breakthrough for brand consistency. Provide a URL to any image and add --sref to apply that image's aesthetic to your generation. This allows you to maintain visual identity across a series of images without writing elaborate style descriptions. Multiple --sref URLs can be combined, each weighted with --sw (style weight, 0-1000).

Character References (--cref)

For consistent character identity across multiple images, use --cref with a character image. V7's character reference is significantly more reliable than earlier versions — the character maintains recognizable features across different poses, settings, and styles. Essential for illustration series, story content, and branded avatars.

Aspect Ratios for Platform-Specific Outputs #

Midjourney supports custom aspect ratios with --ar. Key ratios by platform: --ar 16:9 (YouTube thumbnails, presentations), --ar 9:16 (TikTok, Instagram Reels, Stories), --ar 1:1 (Instagram posts, profile images), --ar 4:5 (Instagram portrait). Setting aspect ratio before generating saves the crop step.

Quality and Stylization Parameters #

--quality controls render time and detail: 0.25 for fast drafts, 1 for standard, 2 for maximum detail. --stylize controls how strongly Midjourney applies its aesthetic training: 0-100 for closer prompt adherence, 100-1000 for increasing artistic interpretation. Finding the right value for your use case takes experimentation — photorealistic work typically uses lower values.

Negative Prompting with --no #

Exclude unwanted elements explicitly: --no text, watermarks; --no blurry, low quality; --no extra limbs (for character work). The --no parameter is more reliable than trying to phrase things as "without X" in the main prompt.

The Remix Workflow #

Enable Remix Mode in settings. When you upscale or create variations, a prompt editor appears — you can modify the prompt while maintaining compositional elements. This is invaluable for iterating on a composition you like while changing style, mood, or specific elements without starting over.

Working with the Web Editor #

The Midjourney web editor allows region-specific editing of generated images — select an area and describe what should replace it. This is the equivalent of Photoshop's generative fill but within the Midjourney aesthetic, meaning edits blend naturally with the surrounding image style.

Professional Output Workflow #

For print-quality work: generate at --q 2, upscale in Midjourney, then run through Magnific AI or Topaz Photo AI for an additional 4x upscale with detail enhancement. This produces files suitable for large-format print. The combination of Midjourney's composition and a dedicated upscaler is currently the professional standard for AI-assisted print work.