Midjourney V8.2 Prompt Guide: Tips and Examples

Midjourney V8.2 rewards clear creative specs—not piles of empty praise words. Since July 24, 2026, V8.2 has been Midjourney’s default model: bolder, more sophisticated house taste, and less willing to invent a full lighting scheme for vague phrases like “beautiful cinematic.” If you still prompt like V7, the default aesthetic often steers the shot.

This Midjourney prompt guide focuses on reusable tactics and copy-paste examples. Paste prompts into Midjourney inside the AI Image Generator, change one variable per round, and keep a short “winner + reject reason” list. For model capabilities, see the Midjourney V8 overview.

What Midjourney V8.2 is especially good at

Reach for Midjourney V8.2 when you need:

1. High-aesthetic exploration — fashion editorials, concept art, brand key-visual drafts
2. Structured prompting — subject, medium, setting, light, and composition stated clearly
3. Style Reference series work — lock look with --sref, then swap only content
4. Personalization continuity — profiles with more ranked images usually read your taste better
5. Native HD finals — roughly 2K when needed; keep early rounds in SD

It is not best at everything. Long in-image copy or rigid product engineering may belong on another model. When you want Midjourney AI art direction with controllable iteration, V8.2 is the current default lane.

A sturdier Midjourney V8.2 prompt skeleton

Write in this order most of the time:

1. Subject — who / what, key identity cues
2. Medium / lane — photography, illustration, concept art, cinematic still
3. Environment — studio, street, interior, seamless backdrop
4. Light + lens — left window light, softbox, 85mm, f/2.0
5. Materials + detail — fabric texture, skin, metal reflections
6. Framing + parameters — crop, negative space, --ar, and when needed --raw / --sref / --no

Keep photography and material terms in English—Midjourney usually tracks them more reliably.

Tip 1: Turn mood words into executable specs

Weak: beautiful cinematic masterpiece, epic mood

Stronger: soft window light from the left, 85mm lens, shallow depth of field, muted teal and cream palette, subtle film grain

V8.2 stops filling in a full house look for you. Skip light and palette, and the bolder default aesthetic will invent one—often not yours.

Tip 2: Use Raw when you need literal following

For complex layouts, product structure, or when the default “editorial bold” look is too loud, add --raw. Migrate old --style raw to --raw.

For series work that already likes V8.2’s edge, leave Raw off for exploration, then tighten on finals.

Tip 3: Change one layer per round

1. Round A — lock subject + light (fast SD batches)
2. Round B — change only wardrobe or only background
3. Round C — change only crop / --ar / negative space
4. Round D — promote winners to HD

Full rewrites every time break the learning loop. Keep a reject-reason note.

Tip 4: Pin Style Reference before content swaps

For brand KVs, character sheets, and series illustration, lock --sref (or a Moodboard) first. --sref carries look, not a photocopy of people or props from the reference.

Tip 5: Set --ar before composition language

Portrait covers, landscape ads, and square ecommerce tiles need the aspect ratio first. Conflicting framing words usually mean --ar was decided too late.

Common starts: --ar 3:4, --ar 16:9, --ar 1:1.

Tip 6: Put refusals in --no

Add concrete exclusions: --no watermark, extra fingers, plastic skin, oversharpened HDR. Vague bad anatomy helps less than specific nouns.

Tip 7: Pick the right tool lane

Goal	Better entry
General Midjourney frames, concepts, fashion, scenes	AI Image Generator
Resume / profile headshots	AI Headshot Generator
Cross-frame character consistency	AI Character Generator
Anime / manga look	AI Anime Generator

Example 1. Portrait with locked light

Use for: brand profile and social identity shots.

Prompt: Half-body portrait of a young woman with natural freckles and visible pores, soft daylight from a large window on the left, gentle shadow under the jaw, off-white wall, face slightly right of center, 85mm lens, f/2.0 shallow depth of field, both ears visible, natural skin texture —ar 3:4 —raw —no beauty blur, plastic skin, heavy makeup, watermark

Rewrite tip: Change age/hair first; keep light + --raw for one full round.

Example 2. Product still with headline room

Use for: ecommerce heroes and ad drafts with right-side layout space.

Prompt: Minimal product still life of a matte white skincare bottle on beige seamless backdrop, softbox key from upper left, gentle bounce fill, crisp bottle silhouette, generous negative space on the right for headline layout, commercial photography —ar 4:5 —stylize 100 —no harsh reflections, extra props, watermark, text

Rewrite tip: Swap bottle color/material only until negative space is stable. Keep any on-bottle words short—or add type later.

Example 3. Fashion editorial locked with Style Reference

Use for: series KVs that must share one look across outfits.

Prompt: Editorial fashion portrait of a woman in a tailored black blazer, cinematic rim light from behind, cool teal fill from the front, sharp catchlights, subtle film grain, medium close-up on 50mm, realistic wool fabric texture —ar 3:4 —sref [your style reference] —stylize 150 —no oversaturated glow, melted facial features

Rewrite tip: After --sref is fixed, change only wardrobe or hair color—not light and reference together.

Example 4. Bold sci-fi concept (lean into V8.2 default edge)

Use for: early concept exploration that benefits from V8.2’s bolder taste.

Prompt: Futuristic canyon city at dusk, neon light strips along cliff roads, volumetric haze, cinematic wide shot, sophisticated sci-fi atmosphere, bold color contrast between amber sky and cyan streets, ultra-detailed architecture silhouettes —ar 16:9 —chaos 10 —stylize 250

Rewrite tip: Raise --chaos while exploring; lower it once direction locks, then consider --raw for structure.

Example 5. Lifestyle interior scene

Use for: home and lifestyle ads that should feel photographed.

Prompt: Woman reading by a large window in a quiet apartment, soft morning side light, linen curtains, light oak table, shallow depth of field, calm lifestyle photography, natural color grading, 35mm documentary feel, realistic interior materials —ar 3:2 —raw —no HDR look, staged smile, plastic skin

Rewrite tip: Name two checkable materials (linen, oak) instead of vague “cozy.”

Example 6. Brand-safe avatar crop

Use for: team pages, resumes, profile drafts.

Prompt: Professional head-and-shoulders portrait of a woman, clean light-gray seamless background, even soft studio light, confident calm expression, centered face with safe crop for circular avatar, natural skin texture, business-casual knit top —ar 1:1 —raw —no heavy retouching, harsh shadows, watermark

Rewrite tip: Draft in Midjourney, then finish in the AI Headshot Generator when needed.

Example 7. Vertical poster with headline room

Use for: event or launch visuals; keep in-image text short or composite later.

Prompt: Vertical poster composition of a confident woman on a golden-hour city street, warm backlight, shallow depth of field, clean upper-third negative space reserved for a short headline, high contrast photographic look —ar 2:3 —stylize 120 —no watermark, busy background clutter, extra slogans

Rewrite tip: Prefer reserved space + design type for long copy; short phrases only for in-image experiments.

Example 8. Neon night portrait with real skin

Use for: music covers and nightlife visuals.

Prompt: Night portrait of a woman under cyan and magenta neon signs, wet street reflections, shallow depth of field, cinematic contrast, rain-damp hair strands, realistic skin under colored light, 50mm lens, moody urban atmosphere —ar 3:4 —raw —no oversaturated glow, melted facial features, plastic skin

Rewrite tip: Swap neon colors for brand accents; keep realistic skin under colored light every round.

Common Midjourney V8.2 prompt mistakes

1. Mood-word spam without light, lens, and palette
2. Changing subject, light, and --sref in one retry
3. Skipping --ar until after composition language is written
4. Burning HD budget on exploration — SD first, HD for winners
5. Heavy edits on HD without re-upscaling — edits can drop back to SD
6. Mixing anime and photo-real cues — pick a lane or use the AI Anime Generator

Next steps

Start with Example 1 or Example 3: copy the skeleton → change one layer → keep reject notes. Pin --sref for series work; use --raw for literal control; promote to HD after direction locks.

Ready to generate? Open the AI Image Generator and turn any prompt above into comparable Midjourney frames—or start generating online. For product overview, stay on the Midjourney V8 model page. Side-by-side craft notes: GPT Image 2 prompt guide and Nano Banana 2 Lite prompt guide.