Midjourney Guide 2025: How to Use, Prompts & Plans
Everything you need to use Midjourney effectively in 2025 — from joining Discord and writing your first prompt, to advanced parameters (--ar, --v 6, --chaos, --seed), the prompt formula that produces professional results, upscaling, /blend, /describe, multi-prompts with ::, and plan pricing from Basic $10/mo to Pro $60/mo.
1. Getting started — Discord setup
Midjourney doesn't have a standalone app — it runs through Discord. Here's how to get started:
Step 1: Go to discord.gg/midjourney and join the Midjourney Discord server (or add the Midjourney Bot to your own server).
Step 2: Subscribe at midjourney.com/account — Basic plan starts at $10/mo. No free trial available since 2023.
Step 3: Generate images in any #general-image-n channel or DM the Midjourney Bot directly for a private workspace.
Step 4: Type /imagine and press Enter, then type your prompt in the text field that appears.
Tip: Use the Midjourney Bot in DMs (/imagine in direct messages) for a cleaner workspace without the noise of public channels.
2. Basic prompts — how /imagine works
The /imagine command generates 4 images from your text description in about 30-60 seconds. Midjourney V6 is the current default model and produces significantly better results than earlier versions.
U buttons (U1-U4): Upscale a specific image — creates a high-resolution version of one of the 4 generated images.
V buttons (V1-V4): Create variations — generates 4 new images similar to the one you selected.
Re-roll (↺): Generate 4 completely new images from the same prompt — different random seed each time.
After upscaling: You see options for Vary (Subtle), Vary (Strong), Zoom Out, and Pan (expand the image in any direction).
Tip: Generate 3-4 batches before upscaling. The first batch rarely produces the best image — re-roll until you see a composition worth refining.
3. Parameters — --ar, --v 6, --style, --q, --chaos, --seed
Parameters go at the end of your prompt and control the technical aspects of generation. You can combine multiple parameters in one prompt.
| Parameter | What it does | Common values |
|---|---|---|
| --ar | Aspect ratio | 16:9, 4:5, 1:1, 2:3, 9:16 |
| --v 6 | Model version (V6 = best in 2025) | --v 6, --niji 6 (anime) |
| --style | Visual interpretation style | raw, expressive, cute, scenic |
| --q | Quality level (time vs detail tradeoff) | --q 1 (fast), --q 2 (best quality) |
| --chaos | Variation randomness (0-100) | 0 (consistent), 50 (varied), 100 (wild) |
| --seed | Lock a specific visual starting point for consistency | --seed 12345 (any number) |
| --no | Exclude elements from the image | --no text, --no people, --no trees |
| --sref | Style reference — match the style of an image URL | --sref [image URL] |
4. The prompt formula — subject + style + lighting + camera
The most consistent way to get professional results is to follow a 4-part prompt structure. Each element gives Midjourney clearer instructions about what to generate.
Style keywords: oil painting, watercolor, digital art, cinematic photography, product photography, street photography, illustration, concept art, ink drawing, film noir
Lighting keywords: golden hour, blue hour, dramatic side lighting, soft diffused light, neon lights, candlelight, studio lighting, overcast, dappled sunlight
Camera keywords: 85mm portrait lens, wide angle, telephoto, macro, Kodak Portra 400, Fujifilm Provia, shallow depth of field, motion blur, 8K ultra-detailed
5. Upscaling and variations
After you generate your initial 4 images, Midjourney gives you powerful tools to refine and enhance your favorites.
Upscale (U1-U4): Creates a high-resolution version (about 1024px) with added detail. Most important step before downloading.
Vary (Subtle): Creates 4 variations that are similar to the upscaled image — small changes while preserving the overall composition.
Vary (Strong): Creates 4 variations with more significant differences — useful when you like the concept but want a different take.
Zoom Out: Extends the image outward — shows more of the scene around the original crop. Options: 1.5x, 2x, or Custom zoom.
Pan (arrows): Expand the image in a specific direction (left, right, up, down) — useful for creating wider compositions from portrait images.
Vary (Region): Select a specific area of the image to regenerate while keeping the rest — useful for fixing a face, changing background, or adjusting one element.
6. Advanced: /blend, /describe, multi-prompts with ::
Three advanced techniques that unlock more creative control.
/blend — combine two images
Type /blend and upload 2-5 images — Midjourney blends their visual styles and content into a new image. No text prompt needed. Useful for style transfer: blend a reference photo with an art style image to apply that style to your content.
/describe — reverse engineer a prompt
Upload any image and type /describe — Midjourney generates 4 text prompts that describe the image's style and content. Use these prompts to generate similar images or understand what keywords produce that aesthetic. Extremely useful for reverse-engineering images you admire.
Multi-prompts with :: — relative prompt weighting
Use :: to separate concepts and assign relative weights. The numbers after :: control how much emphasis each concept gets. Example:
hot dog::2 beach::1 — this gives "hot dog" 2x the weight of "beach", so you'll see a dog that's warm near a beach, not a sausage
Use negative weights (--no or ::-1) to subtract concepts: cat::2 fur::-1 generates a cat but reduces fur-like textures.
7. Plans — Basic $10/mo to Pro $60/mo
| Plan | Price | GPU hours/mo | Generations | Stealth mode |
|---|---|---|---|---|
| Basic | $10/mo | 3.3 fast hours | ~200 images | ✗ |
| Standard | $30/mo | 15 fast hours + unlimited relaxed | ~900+ images | ✗ |
| Pro | $60/mo | 30 fast hours + unlimited relaxed | ~1800+ images | ✓ |
| Mega | $120/mo | 60 fast hours + unlimited relaxed | 3600+ images | ✓ |
Fast mode generates images in 30-60s. Relaxed mode (Standard+) takes 1-10 minutes but uses no fast GPU hours. Stealth mode (Pro+) hides your images from the public gallery.
8. Troubleshooting — stuck jobs, banned content, subscription
Job stuck or not starting? Check midjourney.com/status for outages. In Discord, type /info to see your current job queue. If stuck for 5+ minutes, type /queue to see queue depth, then cancel and re-submit.
Prompt rejected for content? Midjourney moderates for adult content, real people (especially celebrities), trademarked logos, and violent imagery. Rephrase to be more abstract or descriptive of what you want visually without naming restricted subjects.
Fast hours exhausted? Type /info to check remaining fast hours. If you're on Standard or higher, switch to Relaxed mode with /relax — free unlimited generations, just slower (1-10 min). Switch back with /fast.
Subscription not active? Go to midjourney.com/account to verify your subscription. If you subscribed but still see errors in Discord, try leaving and rejoining the server. Check if your Discord account is linked to your Midjourney account.
5 ready-to-use example prompts
Copy and paste these directly into /imagine. Each uses the subject + style + lighting + camera formula with relevant parameters.
a majestic red fox sitting on a snow-covered mountain peak at sunrise, ultra-detailed fur, cinematic lighting, dramatic clouds, --ar 16:9 --v 6 --style raw --q 2
cozy coffee shop interior, morning light through windows, steam rising from espresso cup, warm bokeh, film photography style, Kodak Portra 400 --ar 4:5 --v 6
futuristic city skyline at night, neon lights reflecting on wet streets, cyberpunk aesthetic, ultra-detailed architecture, blade runner 2049 style, cinematic --ar 21:9 --v 6 --chaos 20
portrait of a samurai warrior in feudal Japan, cherry blossom background, ink wash painting style, traditional Japanese art, brush strokes visible --ar 2:3 --v 6 --style expressive
minimalist product photography, white ceramic coffee mug on marble surface, natural side lighting, clean background, commercial photography style --ar 1:1 --v 6 --style raw
Monitor Midjourney status
Midjourney goes down during high-traffic periods — generation queue fills up and jobs get stuck. Track Midjourney uptime at prismix.dev and get free email alerts when it's degraded or down.
FAQ
How do I start using Midjourney?
Midjourney runs through Discord. Go to discord.gg/midjourney, join the server, then subscribe at midjourney.com/account (Basic plan starts at $10/mo). In Discord, type /imagine followed by your prompt in any image channel or in DMs with the Midjourney Bot.
What is the best Midjourney version in 2025?
Midjourney V6 (--v 6) is the default and best version in 2025. Use --v 6 --style raw for photorealistic output. Use --niji 6 for anime and manga styles. V6 produces significantly more detailed and photorealistic results than earlier versions.
What are the most important Midjourney parameters?
--ar sets aspect ratio (--ar 16:9 for widescreen). --v 6 uses the best current model. --style raw makes images more photorealistic. --chaos 0-100 controls variation randomness. --seed locks a visual style for consistency across multiple generations. --no excludes elements from the image.
Is there a free trial for Midjourney?
Midjourney eliminated its free trial in 2023. The cheapest plan is Basic at $10/mo (~200 generations). For free alternatives, try Adobe Firefly (25 free credits/mo), Ideogram (free tier), DALL-E 3 via Bing Image Creator (free), or Leonardo AI (150 tokens/day free).