{"post":{"slug":"midjourney-v7-prompting-guide-2026","title":"Midjourney v7 prompting guide for 2026: style refs, --weird, --chaos, and what changed","excerpt":"v7 changed prompt mechanics meaningfully. Here is the practical 2026 guide to style refs, character consistency, parameter tuning, and the patterns that produce work-ready output.","category":"ai","date":"February 9, 2026","author":"ByteLedger Team","content":"<span class=\"dropcap\">M</span>idjourney v7 launched in mid-2025 and matured through 2026 with major upgrades to style consistency, character consistency, and personalization. The core skill of writing a prompt didn't change, but the parameters changed enough that old habits produce worse results. Here is the current playbook.\n\n## What changed in 2026\n\n- **--sref (style references) and --cref (character references)** are now production-grade. Generate a 20-image series with consistent style and recurring characters.\n- **Personalization (--p) is dramatically better.** With 200+ ratings, Midjourney learns your taste; toggling --p on a prompt biases output meaningfully toward what you've upvoted.\n- **--chaos and --weird** rebalanced — small values (5–25) now produce noticeable variation; large values produce truly weird output that occasionally hits.\n\n## Prompt structure that works\n\nModern Midjourney prompts work best in this order:\n\n```\n[subject], [setting], [composition], [style], [mood/lighting] --params\n```\n\nExample:\n```\na tired barista at 6am, behind a marble espresso bar in a sunlit\nBrooklyn cafe, three-quarter view, photorealistic film photography,\nwarm morning light, hint of steam --ar 3:2 --style raw --v 7\n```\n\n## Style references (--sref)\n\nProvide an image URL or seed; Midjourney transfers style without copying composition. Best use: produce a 10-image series with consistent visual identity.\n\n```\n[your prompt] --sref https://i.mj.run/XXXXX --sw 100\n```\n\n`--sw` controls strength: 0 = ignore style, 1000 = strong style. 100-300 is the sweet spot for most work.\n\n## Character references (--cref)\n\nSame idea for characters. Generate a character once; reuse across images. Best for storyboards, comics, brand mascots.\n\n```\n[your prompt] --cref https://i.mj.run/CHAR --cw 100\n```\n\nStrength `--cw`: 0 = face only, 100 = strong character (face + clothes + hair).\n\n## Personalization\n\nAfter 200 image ratings (Midjourney rolls these into your profile), `--p` biases toward what you've upvoted. Effective for users who consistently rate; less effective if your ratings are inconsistent.\n\n## --chaos and --weird\n\n| Param | Range | What it does |\n| --- | --- | --- |\n| --chaos | 0-100 | Variation across the 4 grid options |\n| --weird | 0-3000 | Aesthetic deviation from \"normal\" |\n| --stylize | 0-1000 | How much Midjourney embellishes |\n\nPractical defaults:\n- Photorealistic: `--chaos 15 --stylize 100 --style raw`\n- Stylized: `--chaos 30 --stylize 500`\n- Exploratory: `--chaos 50 --weird 250`\n\n## Style raw\n\n`--style raw` reduces Midjourney's \"house style\" lean — closer to a photograph and to what you literally described. Default for product shots, photojournalism style, and any time the prompt is precise.\n\n## Negative prompts (--no)\n\nUse for elements you actively don't want. `--no text, watermark, logo, blur` is a useful default for clean output.\n\n## Aspect ratios\n\n`--ar 3:2` for film photography, `--ar 16:9` for YouTube thumbnails, `--ar 9:16` for Reels/TikTok, `--ar 1:1` for Instagram. v7 handles 3:2 and 4:5 better than v6 for portrait subjects.\n\n## Patterns that consistently work\n\n**Character series:** generate a hero character with a strong --cref base, then vary scenarios while keeping --cref + --sref constant.\n\n**Brand asset library:** lock --sref to your brand mood-board image. Generate dozens of consistent visuals.\n\n**Photorealistic product:** `--style raw --stylize 50 --ar 3:2`. Avoid the \"Midjourney glow\" that screams AI on product shots.\n\n**Editorial illustration:** `--stylize 500 --weird 100 --chaos 25` produces the kind of stylized work editors actually buy.\n\n## Common mistakes\n\n**Over-specifying.** Long prompts past 60 words start fighting themselves. Shorter, structured prompts beat long lists.\n\n**Underusing --sref.** Most v6 users haven't switched to using --sref/--cref religiously. The consistency gains are dramatic.\n\n**Ignoring grid choice.** Grid 1-4 of every generation differs. Picking the strongest before upscaling matters more than re-prompting.\n\n## FAQ\n\n**Can I use Midjourney commercially?**\nYes on paid plans (Standard and above). Free trial output is non-commercial.\n\n**Is Midjourney still Discord-only?**\nNo — the web app (alpha.midjourney.com) is now the primary interface in 2026. Discord still works.\n\n**What about Midjourney vs Flux Pro for 2026?**\nMidjourney leads on aesthetic quality and style coherence. Flux Pro leads on prompt adherence and text rendering. Many pros use both.\n\n## Where to go next\n\nFor related guides see [Flux Pro prompting guide](/blog/flux-pro-prompting-guide-2026), [Stable Diffusion 3.5 guide](/blog/stable-diffusion-3-5-guide-2026), and [AI image prompt engineering in 2026](/blog/ai-image-prompt-engineering-2026).","hue":192,"metaTitle":"Midjourney v7 prompting guide (2026): style refs, params","metaDescription":"Practical 2026 prompting guide for Midjourney v7 — style references, character references, --chaos, --weird, personalization, and proven patterns.","keywords":["midjourney v7 guide","midjourney prompting 2026","midjourney style reference","midjourney character reference","sref cref midjourney","midjourney chaos weird","midjourney personalization","midjourney parameters","ai image prompting","midjourney best practices"],"tldr":["**--sref + --cref are the productivity unlock.** Style and character consistency across a series.","**--chaos and --weird are blunter in v7** — small values produce visible variety.","**Personalization (--p) has matured** — train Midjourney on your taste with 200+ ratings.","**Skip:** treating Midjourney as a Stable-Diffusion replacement. Different strengths — pick by job."],"readTime":"3 min read","wordCount":684},"related":[{"slug":"how-to-use-midjourney-v8-2026","title":"How to use Midjourney v8 in 2026: prompts, parameters, and real results","excerpt":"Midjourney v8 raised the ceiling on photorealism and controllability. This guide covers the prompts, parameters, and workflows that get professional results fast.","category":"ai","date":"February 28, 2026","author":"ByteLedger Team","readTime":"5 min read","hue":214},{"slug":"how-to-use-midjourney-2026","title":"How to use Midjourney in 2026: a beginner guide to AI images","excerpt":"Sign up, write a clear prompt describing subject, style, and lighting, then refine with variations and parameters. Here is the full beginner workflow.","category":"ai","date":"June 8, 2026","author":"ByteLedger Team","readTime":"3 min read","hue":305},{"slug":"how-to-use-sora-2026","title":"How to use Sora in 2026: video generation, prompting, and what it actually does well","excerpt":"Sora moved from waitlist to widely available in 2026, but most users still misuse it. Here is what it does well, its real limits, and how to prompt for usable video.","category":"ai","date":"March 18, 2026","author":"ByteLedger Team","readTime":"4 min read","hue":208}],"sameCategory":[{"slug":"agent-cost-attribution-2026","title":"Agent Cost Attribution: Finding Where the Tokens Went","excerpt":"A single monthly API bill tells you that agents are expensive. It does not tell you which feature, which customer, or which retry loop is responsible. That gap is fixable.","category":"ai","date":"August 19, 2026","author":"ByteLedger Team","readTime":"5 min read","hue":220},{"slug":"human-in-the-loop-agents-2026","title":"Human-in-the-Loop Agents: Where to Put the Approval Gate","excerpt":"Approve everything and the agent is a slow autocomplete. Approve nothing and you have handed irreversible actions to a system that can be talked into things.","category":"ai","date":"August 19, 2026","author":"ByteLedger Team","readTime":"5 min read","hue":197},{"slug":"reasoning-effort-controls-2026","title":"Reasoning Effort Controls: Tuning How Hard a Model Thinks","excerpt":"Fixed thinking budgets gave way to effort levels, and the default is rarely the right setting. Here is how the dial behaves and where each level actually belongs.","category":"ai","date":"August 19, 2026","author":"ByteLedger Team","readTime":"5 min read","hue":210},{"slug":"retrieval-metrics-explained-2026","title":"Retrieval Metrics: Measuring the Half of RAG That Fails","excerpt":"When a RAG system answers badly, the retrieval usually failed before the model ever spoke. Measuring generation quality alone hides that entirely.","category":"ai","date":"August 19, 2026","author":"ByteLedger Team","readTime":"5 min read","hue":215}],"newer":{"slug":"local-llm-setup-guide-2026","title":"Local LLM setup guide for 2026: hardware, quantization, when to bother","excerpt":"Running LLMs locally got dramatically more practical in 2026. Here is the honest guide to hardware, quantization, model selection, and whether self-hosting actually makes sense for you.","category":"ai","date":"February 9, 2026","author":"ByteLedger Team","readTime":"4 min read","hue":196},"older":{"slug":"neon-vs-supabase-2026","title":"Neon vs Supabase in 2026: Postgres-as-a-service compared","excerpt":"Branching, edge compatibility, pricing, and DX — the honest 2026 comparison of Neon and Supabase, with the project type each one wins.","category":"code","date":"February 9, 2026","author":"ByteLedger Team","readTime":"3 min read","hue":180}}