Midjourney v7: Create Professional Images with AI in 15 Minutes

Difficulty: Intermediate

Category: Ai Tools

Midjourney v7: Create Professional Images with AI in 15 Minutes

By March 2026, 73% of marketing teams use AI-generated images in their campaigns, according to the Content Marketing Institute. Midjourney v7, released in January 2026, has become the industry standard because it finally solves the consistency problem—you can now generate brand-aligned images with repeatable style codes and character references that previously required a $5,000/day photographer.

This tutorial will teach you how to create professional-grade images using Midjourney v7’s most powerful features: style reference codes, consistent characters, advanced parameters, and the new --quality flag that dramatically improves output.

Prerequisites

Before starting, make sure you have:

• Active Midjourney subscription ($10/month Basic or $30/month Standard—you’ll need Standard for stealth mode and more generations)
• Discord account connected to Midjourney bot (v7 is Discord-based until the web interface launches Q2 2026)
• Basic prompt writing experience (you should understand what “/imagine” does)
• Reference images ready if you want to use style references (optional but recommended)

Step 1: Access Midjourney v7 and Set Your Default Version

Midjourney doesn’t automatically use v7—you need to activate it explicitly.

In any Midjourney Discord channel, type:

/settings

A control panel appears. Click the “MJ Version 7” button (it turns green when active). This sets v7 as your default for all future prompts.

Alternatively, add the version flag to individual prompts:

/imagine prompt: professional headshot of a female CEO, natural lighting --v 7

Gotcha: If you forget to set v7, Midjourney defaults to v6.1, which produces noticeably different results. Always check your settings panel shows “7.0” in green.

Pro Tip: Use /settings to also enable “Remix mode” (lets you modify prompts during upscaling) and “High variation mode” (increases diversity between the 4 image grid results).

Step 2: Master the New Quality Parameter

Midjourney v7 introduces --quality (or --q) with granular control from 0.25 to 2.0. This replaced the binary quality system in v6.

Quality values and when to use them:

• --q 0.25 → Fast drafts, costs 0.25 GPU minutes (use for iteration)
• --q 1 → Standard quality, costs 1 GPU minute (default, balanced)
• --q 1.5 → Enhanced detail, costs 1.5 GPU minutes (product photography)
• --q 2 → Maximum fidelity, costs 2 GPU minutes (hero images, print work)

Example prompt with quality control:

/imagine prompt: macro shot of water droplets on green leaf, morning dew, shallow depth of field --q 1.5 --ar 16:9

This generates a 16:9 image with enhanced detail perfect for a website hero banner.

Gotcha: Higher quality doesn’t mean “better” images—it means more rendering time and detail. A --q 2 portrait might show pores and skin texture you don’t want. Test at --q 1 first.

Step 3: Use Style Reference Codes for Brand Consistency

This is v7’s killer feature. Style reference codes (--sref) let you maintain visual consistency across dozens of images.

How it works:

1. Generate an image you love with the style you want
2. Copy its job ID (the code after the filename, like 2847392847_384729)
3. Use --sref in new prompts to match that style

Real example workflow:

/imagine prompt: minimalist product photography, white background, soft shadows --q 1

You get 4 images. The third one (job ID: 1234567890_abcdef) has the perfect clean aesthetic. Now generate 10 more products in that exact style:

/imagine prompt: wireless headphones on white surface --sref 1234567890_abcdef
/imagine prompt: smart watch with leather band --sref 1234567890_abcdef
/imagine prompt: portable bluetooth speaker --sref 1234567890_abcdef

All three images will match the lighting, composition style, and color grading of your reference.

Pro Tip: Combine multiple style references with weights: --sref 1234567890_abcdef::0.7 9876543210_fedcba::0.3 blends two styles (70% first, 30% second).

Step 4: Generate Consistent Characters with Character Reference

Need the same person across multiple images? Use --cref (character reference).

Step-by-step process:

1. Generate your ideal character:

/imagine prompt: professional portrait of a 30-year-old asian woman, short black hair, glasses, confident expression, studio lighting --q 1.5

1. Upscale your favorite version (click U1, U2, U3, or U4 under the grid)

2. Copy the image URL (right-click the upscaled image → Copy Link)

3. Use that URL in future prompts:

/imagine prompt: [paste image URL] professional woman giving a presentation, conference room, gesturing at screen --cref [same image URL] --cw 100

The --cw (character weight) parameter ranges from 0-100:

• --cw 0 = only face match
• --cw 50 = face + approximate outfit
• --cw 100 = exact face, hair, and clothing match

Gotcha: Character reference works best with clear, front-facing portraits. Side angles or obscured faces reduce consistency by 40-60%.

Step 5: Fine-Tune with Advanced Parameters

Midjourney v7 adds precision controls that professionals need:

Aspect Ratio (--ar):

• --ar 1:1 → Instagram posts
• --ar 4:5 → Instagram portraits
• --ar 16:9 → YouTube thumbnails, website heroes
• --ar 9:16 → Mobile screens, Stories
• --ar 3:2 → Standard photography ratio

Stylization (--s): Range 0-1000 (default is 100)

• --s 0 → Literal interpretation, photorealistic
• --s 100 → Balanced artistic interpretation
• --s 750 → Highly artistic, painterly

Chaos (--c): Range 0-100 (default is 0)

• --c 0 → All 4 grid results are similar
• --c 50 → Moderate variation
• --c 100 → Wildly different interpretations

Real-world example for client work:

/imagine prompt: luxury hotel lobby, marble floors, gold accents, chandelier, afternoon sunlight through tall windows --ar 16:9 --s 50 --q 1.5 --c 10

This generates a photorealistic architectural shot (--s 50 keeps it grounded) with slight variation (--c 10) so you have options, in a web-ready format.

Step 6: Use Negative Prompts Effectively

Tell Midjourney what to avoid with --no:

/imagine prompt: portrait of elderly man, natural wrinkles, wisdom --no smooth skin, beauty filter, artificial, plastic

Common negative terms that improve results:

• --no text, watermark, signature (removes unwanted text)
• --no blurry, out of focus (increases sharpness)
• --no oversaturated, HDR (prevents that “AI look”)
• --no multiple heads, distorted hands (reduces anatomical errors)

Pro Tip: v7 significantly improved hand and text rendering, but complex hand positions still fail 20-30% of the time. Use --no distorted hands and generate 8-12 variations.

Step 7: Upscale and Edit with Vary Region

After generating your 4-image grid, you have new options:

Upscale buttons (U1-U4): Creates a 2048x2048 (or larger with --q 2) final image

Vary buttons (V1-V4): Generates 4 new variations based on that specific image

NEW in v7: Vary (Region):

1. Click “Vary (Region)” under any upscaled image
2. A selection tool appears—paint over the area you want to change
3. Type what you want in that region: “holding a coffee mug” or “sunset background”
4. Midjourney regenerates only that area

This is like Photoshop’s generative fill but inside Discord.

Example use case: You generated a perfect portrait but the background is boring. Click Vary (Region), select the background, type “blurred city skyline at dusk” and regenerate.

Practical Example: Creating a Complete Brand Identity

Here’s a real workflow for creating consistent marketing images for a fictional coffee brand:

Step 1 - Establish the style:

/imagine prompt: artisanal coffee cup on rustic wooden table, warm morning light, shallow depth of field, cozy atmosphere, steam rising --ar 4:5 --q 1.5 --s 75

Save your favorite as style reference: --sref 8372649201_ac82bd

Step 2 - Generate product variations:

/imagine prompt: pour over coffee brewing, close-up shot --sref 8372649201_ac82bd --ar 4:5
/imagine prompt: coffee beans scattered on wooden surface --sref 8372649201_ac82bd --ar 4:5
/imagine prompt: barista hands creating latte art --sref 8372649201_ac82bd --ar 4:5

Step 3 - Add consistent person:

/imagine prompt: friendly barista portrait, apron, smiling, natural light --sref 8372649201_ac82bd --q 1.5

Use this upscaled image URL as --cref for all staff photos.

Step 4 - Create location shots:

/imagine prompt: cozy coffee shop interior, customers chatting, [barista URL] working behind counter --cref [barista URL] --sref 8372649201_ac82bd --ar 16:9

Result: 12-15 images in 20 minutes that look like they came from the same professional photoshoot. Total cost: ~$0.40 in GPU minutes with a Standard plan.

Key Takeaways

• Midjourney v7 requires explicit activation via /settings or --v 7 flag—it’s not automatic
• Style reference codes (--sref) are game-changers for brand consistency, saving hours of iteration and maintaining visual coherence across campaigns
• The --quality parameter (0.25-2.0) directly controls cost and detail—use --q 0.25 for rapid iteration, --q 1.5 for final outputs
• Character reference (--cref) with --cw 100 creates repeatable people across multiple scenes, finally solving the “same person” problem that plagued earlier versions

What’s Next

Once you’ve mastered v7’s core features, explore the Midjourney API (beta access rolling out April 2026) to automate image generation directly from your applications without touching Discord.

---

Key Takeaway: Midjourney v7 introduces style reference codes, consistent character generation, and advanced parameter control that let you create production-ready images 3x faster than v6, with fine-tuned control over lighting, composition, and brand consistency.

---

New AI tutorials published daily on AtlasSignal. Follow @AtlasSignalDesk for more.

---

📧 Get Daily AI & Macro Intelligence

Stay ahead of market-moving news, emerging tech, and global shifts.