Midjourney Styles & Scenes Playbook

Updated: 2026-08-29. UI options, personalization, and parameter syntax follow midjourney.com and official docs that day. Midjourney has no official public API; third-party entries ≠ official service.

Introduction

Searches for “Midjourney styles,” “scene playbook,” or “parameter practice” usually mean a team wants reusable style cards from scattered wins—not a blank dialog every time. This page covers high-frequency scenes—posters, covers, portraits, products—with constraint focus, prompt skeletons, and acceptance criteria. Prompt structure pairs with the Prompt playbook; grid ops pair with Getting started.

What this guide solves

• Pick photography / illustration / 3D / infographic and a model track from a style matrix
• Get four style cards and skeletons: poster, cover, portrait, product
• Steady series vibe with personalization / moodboards / --sref (official that day)
• Generate from scene recipes and ship with one shared checklist

Style matrix: use first, look second

Style lean	Better for	Nail in the prompt	Use carefully	Model lean
Commercial photography	Ecommerce mains, physical display	Material, contact shadow, clean background	Hard fictional worlds	General (e.g. V8.2)
Editorial photography	Blog heroes, brand stories	Lens feel, mood light, whitespace	Precision part close-ups	General
Flat / vector illustration	Social, explainer art	Palette, stroke, icon geometry	Real skin texture	General or Niji
Anime / thick-paint illustration	Characters, sheets, manga covers	Line art, coloring method, costume layers	ID-grade real people	Niji 7
3D render	Concept products, game assets	Material spheres, env light, clean edges	ID-grade portraits	General
Infographic	Process, comparison, tip posters	Zones, short titles, safe areas	Dense data tables	General + post type

Selection rule: ask what communication the frame must finish, then pick style and track—don’t stack empty “ultra detail, 8K, masterpiece” words. V8.2 depth: V8.2 guide.

One frame, one communication job

Trying to sell + tell a brand story + dump campaign info + deliver portrait emotion in one frame almost always fails. Before writing, ask:

• In a feed, what must be recognizable in 3 seconds?
• On open, should the first eye land on subject, title zone, or product?
• Is copy generated in-frame or overlaid in post?

If those three are unclear, don’t generate yet. Answer them, pick one matrix row, open the matching style card. That cuts “flashy style, unusable on channel” rejects.

How to write a style card (team template)

Keep each card ≤15 lines with fixed fields:

Style card name: ________
Model track: general / Niji (UI name that day)
Channel / use: ________
Aspect: --ar __ (or official that day)
One-line primary style: ________
Palette: primary / secondary / banned
One-line light: ________
Must keep: ________
Bans: ________
Master / sref ID: ________
Winning prompt summary: ________
Failure notes: ________

New briefs start from a style card, then change subject and copy—don’t rewrite the whole novel.

Scene 1: posters

Goal: theme readable from afar, layout-ready up close, title safe zone reserved.

Constraint focus:

• Lock aspect (1:1 / 4:5 / 16:9) and which side holds text
• Prefer title overlays in post over forced generation
• Brand colors as concrete names or HEX intent—not “nice colors”
• High --stylize often looks pretty but disobeys—finals can sit slightly lower (official that day)

Skeleton:

4:5 campaign poster base; top 40% whitespace for title; lower subject [description];
[flat illustration / editorial photo]; primary [color] secondary [color];
no on-image text; ban microcopy, QR codes, extra logos, watermarks.
--ar 4:5

Scene 2: covers (blog / social / album feel)

Goal: thumbnail-recognizable, atmospheric at full size, subject intact after crop.

Constraint focus:

• Landscape 16:9 reserves a title zone; squares need a center safe zone
• Few subjects (one primary visual)—thumbnails otherwise mush
• Mood via light and color temperature—not abstract adjective piles

Skeleton:

16:9 article cover; subject on the right [one line], ~30% left whitespace;
shallow depth, [morning window light / neon rim]; editorial photography;
no text no watermarks; no extra people stealing focus.
--ar 16:9

Album / playlist squares: switch to 1:1 and write “centered composition, edges croppable.”

Thumbnail stress test

Shrink the frame to feed size (about a thumb wide):

1. Can you still recognize the subject?
2. Did large color blocks turn muddy gray?
3. If there is on-image type, is it still readable? (Usually no → proof you should overlay in post)

Covers that fail the thumbnail test rarely finish the “click” job no matter how pretty. Return to the style card: fewer subjects, stronger value contrast—not more detail words.

Scene 3: portraits

Goal: same person recognizable across frames, or one frame with accurate mood and intact features.

Constraint focus:

• First pass lock: age feel, hair, costume color, signature accessory
• Later: “keep the same person identity,” with character refs when possible (parameter names follow official that day)
• Change only scene or action—don’t rewrite hair, age feel, and ethnicity cues together
• When hands enter frame: “five fingers clear” or crop to bust

Skeleton:

Bust portrait, eye level, soft side light; [age feel] [hair] [costume color];
expression [keyword]; clean soft background; natural skin—not plastic over-smooth;
ban: extra fingers, watermarks, on-image text, unauthorized brand logos.
--ar 3:4

Anime character portraits: switch to Niji and use line/color language—see Niji 7.

Portrait failure → next action

Symptom	Prefer next
New face every frame	Fixed identity sentence + character ref; change scene only
Hands collapse	Bust crop or “hands out of frame / five fingers clear”
Over-smooth plastic	“Natural skin texture, not plastic”
Random clothing brands	Ban unauthorized logos; describe solid-color garments

Scene 4: products

Goal: believable materials, complete subject, croppable into detail pages or multi-angle series.

Constraint focus:

• State product geometry and must-keep color blocks / label placement
• Nail background in one line (solid, gradient, or scene)—don’t let the model invent props
• Multi-angle: fix light and background; change only “camera position”
• Microcopy, barcodes, complex logos: safer in post

Skeleton:

1:1 ecommerce main; [product] 3/4 angle fully in frame; seamless light-gray background;
soft top-side light; real [material] texture; natural contact shadow;
no text no watermarks; no hands or extra props.
--ar 1:1

Lifestyle “hand holding product”: lock product silhouette on a clean main first, then open a separate lifestyle scene—lowers fused fingers and warps.

Fixed sentences for multi-angle series

For front / side / top-down, copy the same style card and only replace the camera line:

Keep the same product, same light-gray seamless background, same top-side light and material description.
Only change camera position to: [front eye-level / 3/4 / slight 45° top-down].
Product fully in frame; no new props; no text or watermarks.

Accept each angle on its own grid—don’t expect one prompt to spit a “nine-cell manual.” When silhouette must match packaging engineering drawings, upload an object ref and write “geometry follows the reference” (parameter names follow official that day).

One-line acceptance per scene

Shrink acceptance to one line at the top of the brief—reviews move faster:

Scene	One-line acceptance
Poster	Thumbnail reads the theme; title zone has enough blank
Cover	Subject survives crop; mood matches the article
Portrait	Features intact; identity matches last week’s lock
Product	Materials believable; silhouette matches pack source

When debate is “pretty or not,” return to that one line first—then argue taste.

Personalization, moodboards, and series vibe

How to use a Moodboard

1. Collect 4–8 rights-cleared refs: palette and light only—not full living-artist clones.
2. Distill three sentences: color temperature, material, camera distance.
3. Pick one vibe-defining finished frame as --sref master (syntax per official that day).
4. New prompts change story only—not those three vibe lines.

Personalization

If the official product offers personalization / profile features (names follow that day):

• Fits long-term personal taste; don’t blindly share one profile across every client brand
• Brand projects: prefer style card + sref; treat personalization as optional acceleration—not the only source
• After account or entry switches, personalization may not migrate—especially on third parties

Series recipe

Same style card + same aspect + same bans
× change only: scene / action / SKU color
= a reviewable batch

Log fields: date | entry | model | style-card name | aspect | prompt summary | sref/cref | selected?.

Three-sentence moodboard review rule

Showing moodboards to non-designers: never dump nine images alone. Speak or write three lines:

1. Light we want: e.g. “window cool/warm contrast; highlights not blown.”
2. Materials we want: e.g. “matte ceramic + light wood—not metal high gloss.”
3. Vibe we refuse: e.g. “no cyber, no horror, no over-smooth skin.”

After those three pass, batch-generate in Midjourney. Otherwise review becomes “I like the top-left one” taste voting and the style card never freezes.

Scene recipe cheat sheet

Scene	Aspect start	One style line	Parameter note (that day)
Campaign poster	4:5 / 3:4	Flat or large-block illustration	Whitespace; text in post
Blog cover	16:9	Editorial photography	Single subject; don’t max --stylize blind
People social	3:4 / 4:5	Natural-light portrait	Lock identity; careful with features
Ecommerce main	1:1	Commercial product photography	Nail background; material first
Character sheet	2:3	Clean line-art illustration	Niji track; complete hands/feet
Concept 3D	16:9 / 1:1	Clean material-sphere render	Fewer messy scene lights

Deeper parameters and negatives: Prompt playbook.

Combo scene: three-image landing set

Many landings need “hero KV + benefit illustration + product close-up” as a set. Suggested flow:

1. Freeze shared palette and bans as one mother style card.
2. Hero KV uses the cover skeleton (16:9, large whitespace).
3. Benefit art uses poster/flat skeleton (e.g. 4:5); subject becomes iconized scene; same palette.
4. Product close-up uses the product skeleton (1:1); light/background echoes the KV without forcing the same composition.
5. All three pass the checklist, then unify fonts and logos in post.

Don’t try one prompt that “generates the whole page”—controllability collapses. Separate frames, same card, unify in post: that’s a deliverable set.

How teams capture style cards

1. Every winning frame saves: final prompt, model track, aspect, ref/master ID.
2. Extract reusable sentences into a style card (≤15 lines); ban private adjective dialects.
3. New briefs start from the card, then change subject and copy.
4. Quarterly review: delete outdated catchphrases and dead parameters (check official docs that day).
5. Archive by track: don’t mix general photo masters with Niji illustration masters.

Minimum style-card review flow

1. Designer submits: style card + 3 win samples + 1 fail sample (with cause).
2. Stakeholders judge channel fit and brand color only—not “I personally prefer another light.”
3. On pass, warehouse it; on fail, edit card fields only—don’t rewrite every historical frame that day.
4. Big promo or season change: allow a “seasonal variant card” that still inherits bans and aspect from the mother card.

Keep fail samples—they tell newcomers “this path is closed,” which beats another thousand adjectives. Parameter and template depth: Prompt playbook; ops loop: Getting started.

Completion checklist

• Style matches use (matrix chosen); model track correct
• Aspect and safe zones match channel specs
• Subject count, hands, edge artifacts checked
• On-image text proofed or moved to post
• No surprise brand logos / watermarks
• Portrait/character cross-frame consistency acceptable (when required)
• Style card and master ID warehoused
• Third-party privacy noted; no assumed official public API

Quick access

• China-friendly (third-party): Midjourney V7
• Mirror studio (third-party): Text-to-image studio
• Official: www.midjourney.com

Accounts, quotas, and model lists follow each entry that day; prices are not hard-coded here. China paths: China access guide.

FAQ

Style card vs prompt template?

Templates are generic skeletons; style cards are verified frozen brand/project aesthetics (palette, master, bans). Templates first; promote winners to style cards.

Why does the same prompt feel different today vs yesterday?

Model version, personalization state, parameter defaults, even entry routing can change. Notes need “UI model name + date” or comparisons are meaningless.

Must posters generate the title in-frame?

No. Most delivery is steadier as: Midjourney no-text base → design tool overlays titles and logos.

Is --weird good for style exploration?

Good for moodboard divergence; bad for spec ecommerce mains. Explore on; for finals, turn off or lower to docs’ suggested low range (official that day).

Prices?

Follow official or third-party pages that day; this article does not hard-code unit prices.

Official resources

• midjourney.com
• Official help and docs entry (site that day)

Further reading

• Midjourney prompt playbook
• Midjourney getting started: first image
• Midjourney V8.2 full guide
• Midjourney Niji 7 selection & practice
• Midjourney Guides Hub
• What is Midjourney? Family map
• Midjourney China access guide

Summary

Styles & scenes come down to: use → matrix for style and track → freeze vibe in a style card → generate from scene skeletons → ship with the checklist. Posters and covers care about safe zones; portraits care about identity lock; products care about materials and background. Personalization and --sref steady series—they don’t replace acceptance sheets. Parameters and feature names follow midjourney.com that day; third-party ≠ official, and there is no official public API.