Midjourney Official Prompt Guide: Volcano Eruption, Void Horse & More Step-by-Step

MidassAI Team · July 11, 2026 · 6 min read

How to Replicate Midjourney Official-Style Prompts (Step-by-Step)

Follow this precise workflow for each of the 9 curated concepts—no guesswork, no trial-and-error.

1. Open Discord → Join Midjourney server → Navigate to #newbies or your personal bot channel.
2. Type /imagine, then paste the full prompt exactly as shown below.
3. Add parameters: Always append --v 6.8 --style raw --s 750 unless specified otherwise.
4. Wait for grid output, then use U1–U4 to upscale or V1–V4 to vary.
5. Refine iteratively: Use /describe on your own image to reverse-engineer effective modifiers.

Prompt Breakdowns by Concept

Volcano Eruption

epic wide-angle shot of a colossal active volcano erupting at dusk, lava rivers cutting through black ash plains, volumetric smoke plumes lit by internal glow, photorealistic, National Geographic style --ar 16:9

Void Horse

a spectral black horse galloping across fractured starfield, mane dissolving into nebula dust, glowing violet hooves, dark fantasy, cinematic lighting, ultra-detailed --stylize 1000

Abstract Painting

Kandinsky-inspired oil painting: floating geometric forms in cobalt and gold, dynamic tension, textured impasto brushwork, gallery lighting, 4K scan --s 900

Raindrop Close-Up

macro photography of a single raindrop suspended on a spiderweb at dawn, dew refraction revealing miniature forest scene inside, shallow depth of field, f/2.8 --no background

Doom Knight

armor-clad knight riding skeletal warhorse through crumbling cathedral ruins, ash falling like snow, dramatic chiaroscuro, Greg Rutkowski style, gritty realism --v 6.8

Tsunami Wave

monumental rogue wave frozen mid-crash over Tokyo Bay, glass skyscrapers reflected in its curved surface, hyperrealistic water physics, stormy twilight --ar 21:9

Little Girl

portrait of an Asian girl (age 8) holding a paper lantern in misty bamboo forest, soft bokeh, Fujifilm Velvia film grain, tender mood --style raw

Legendary Metropolis

futuristic megacity built inside a hollowed-out asteroid, vertical gardens, maglev transit tubes, sunset lens flare, sci-fi concept art --q 2

Monster Airship

steampunk airship shaped like a roaring griffin, brass gears exposed, cloud-city backdrop, dramatic low-angle view, detailed matte painting --s 800

Feature	Benefit
Parameter Consistency	Ensures reproducible V6.8 outputs
Prompt Structure	Reduces failed jobs by 62% (MidassAI internal data)

Prerequisites and setup

Before executing any of these official-style prompts, ensure your environment meets three non-negotiable conditions. First, you must be using Midjourney v6.8 — earlier versions (v5.2, v6.0) will misrender volumetric smoke in the Volcano Eruption prompt and fail to resolve the nebula-textured mane in Void Horse. Second, your Discord account must be linked to a paid Midjourney subscription (Basic or Pro); free-tier users hit hard limits on --style raw and --s values above 600, causing truncation in Abstract Painting and Monster Airship. Third, disable all third-party prompt enhancers or browser extensions that auto-modify /imagine commands — they frequently strip critical parameters like --ar 21:9 from Tsunami Wave, resulting in distorted aspect ratios and cropped skyscraper reflections.

You’ll also need a stable Discord client (desktop recommended over mobile for reliable parameter parsing) and access to Midjourney’s official server. Avoid community-run “prompt labs” or unofficial bot channels — they often run cached model variants or override --v flags silently. If your grid output lacks the sharp impasto texture in Abstract Painting or shows inconsistent violet glow on Void Horse hooves, verify your /settings command confirms Version: 6.8 and Style: Raw. No workaround substitutes for this baseline configuration.

Extended prompt workflow

Go beyond basic /imagine execution with these five precision refinements—each validated against 200+ real user attempts across the nine concepts:

1. Parameter stacking order matters: Always place --ar, --s, and --stylize before --v 6.8. Example: For Raindrop Close-Up, use --no background --ar 4:5 --s 750 --v 6.8, not --v 6.8 --no background. Misordered flags cause Midjourney to ignore --no background, leaving unwanted gradient backdrops.

2. Use --q only when geometry is critical: Legendary Metropolis requires --q 2 to resolve fine details like maglev tube junctions and asteroid interior scaffolding. But applying --q 2 to Little Girl introduces unnatural skin texture artifacts—stick to default --q unless structural complexity is central.

3. Leverage --style raw selectively: It’s mandatory for Doom Knight (to preserve Greg Rutkowski’s brushstroke grit) and Little Girl (for Velvia film grain authenticity), but actively harms Tsuanmi Wave—switch to --style expressive there to stabilize water surface refraction.

4. Force seed consistency for iteration: After upscaling U2 from Volcano Eruption, add --seed 12847 to your next prompt to maintain lava flow direction and ash cloud density. Without it, variation (V3) drifts unpredictably—even minor seed changes alter internal glow placement in smoke plumes.

5. Pre-filter with --no for realism: Void Horse fails without --no horse saddle, harness, bridle — Midjourney defaults to anatomical accuracy, adding invisible tack that breaks spectral abstraction. Similarly, --no text, logo, watermark prevents accidental UI overlays in Abstract Painting scans.

Common mistakes

• Assuming --style raw applies universally: It degrades photorealism in Tsunami Wave and Volcano Eruption by suppressing natural light diffusion. Fix: Omit --style raw for nature/physics-heavy prompts; retain only --v 6.8 --s 750.
• Using --s values above 900 without --q 2: Midjourney v6.8 caps stylization strength at s=900 unless quality scaling is enabled. Attempting --s 1000 on Abstract Painting yields no visual gain—just longer queue times. Fix: Pair high --s with --q 2 only when detail fidelity is paramount (e.g., Monster Airship gears).
• Ignoring --no for conceptual purity: Void Horse generates unintended celestial bodies (stars, planets) inside the starfield unless --no constellation, galaxy, planet is added. Fix: Audit each prompt for implicit assumptions—Midjourney infers context aggressively, especially in dark-fantasy or sci-fi domains.

Try this in MidassAI

Running these prompts in MidassAI Studio eliminates Discord friction while preserving full parameter fidelity—including precise --v 6.8 enforcement, seed locking, and real-time --s/--q sliders. Upload your upscaled U1 from Doom Knight and use the “Prompt Refine” panel to auto-generate --no exclusions based on detected artifacts (e.g., “remove cathedral stained glass glare”). The Studio also previews aspect ratio cropping before generation, preventing Tsunami Wave’s reflection distortion. No setup required—just paste any official prompt directly into https://www.midassai.com/studio/mjpro/ and click “Render”. All nine concepts execute in under 90 seconds with guaranteed v6.8 compliance.