Listen to this article
Midjourney is an AI tool for generating images and videos from text prompts, accessible primarily through its web interface or Discord, with subscription plans starting at $10/month for basic use.
Midjourney is a powerful AI driven platform that transforms text descriptions into stunning images and short videos, leveraging advanced diffusion models to create visuals ranging from realistic photographs to abstract art. It has gained popularity among artists, designers, and hobbyists for its artistic quality and iterative features. As of early 2026, Midjourney operates on models like V7 (released in April 2025), with rumors of V8 enhancements focusing on UI overhauls, improved style references, and editing tools potentially rolling out around January.
Step 1: Subscription and Account Setup
To use Midjourney, a paid subscription is required, as free trials are no longer available. Plans are structured around GPU time for image generation, with “Fast” mode using allotted hours (about 60 seconds per generation) and “Relax” mode offering unlimited but slower processing on higher tiers.
| Plan | Monthly Cost (Annual Discount) | Fast GPU Hours | Key Features | Best For |
| Basic | $10 ($8 annually) | 3.3 hours (~200 images) | Basic access, personal use | Hobbyists, casual beginners |
| Standard | $30 ($24 annually) | 15 hours + unlimited Relax | General queue, concurrent jobs | Regular creators, designers |
| Pro | $60 ($48 annually) | 30 hours + unlimited Relax | Stealth Mode (private generations), 12 concurrent jobs | Professionals, privacy focused users |
| Mega | $120 ($96 annually) | 60 hours + unlimited Relax | All Pro features, 12 concurrent jobs | Heavy users, agencies, high volume needs |
Subscribe by logging into midjourney.com, selecting a plan, and completing payment. Ownership of images is granted to paid users for personal/commercial use, but companies earning over $1M/year need Pro or Mega. Create a Discord account at discord.com if you don’t have one, as it’s required for authentication, even if using the web app.
Step 2: Choosing Your Workflow: Web vs. Discord
Midjourney supports two main interfaces, with the web recommended for beginners due to its intuitive design and organization tools.
- Web Workflow: Navigate to https://www.midjourney.com/imagine. This provides a dedicated “Create Page” with an Imagine bar for prompts, real time progress bars (showing generation up to 100%), and easy access to settings. Benefits include drag and drop image uploads for references, built-in organization on the “Organize Page,” and collaboration via “Chat Page” rooms. It’s ideal for solo work without public distractions.
- Discord Workflow: After subscribing, click “Join the Beta” on midjourney.com to enter the official server. Use channels like #newbies for public generations, or create a private server by inviting the Midjourney Bot (search for it in Discord and add to a new server). Commands are bot-based, starting with / (slash). This is great for community feedback but can be noisy for beginners.
For either, customize default settings via the gear icon (web) or /settings (Discord), adjusting aspects like model version (e.g., V6 for consistency or V7 for improved prompts), stylize level, and GPU speed.
Step 3: Crafting Effective Prompts
Prompts are the core of Midjourney, natural language descriptions that guide the AI. Start with basics and build complexity to avoid overwhelming the model. A good prompt structure includes: subject + attributes + scene + style + technical details.
Beginner Prompt Examples:
- Simple: “a cat”
- Improved: “a fluffy orange tabby cat sitting on a windowsill, afternoon sunlight, cozy atmosphere”
- Advanced: “a fluffy orange tabby cat sitting on a Victorian windowsill, golden hour light streaming through lace curtains, dust particles visible in the light, oil painting style, warm color palette, intimate atmosphere –ar 3:4 –s 750”
Tips: Be descriptive but concise (under 75 words ideally); reference artists/styles (e.g., “in the style of Studio Ghibli”); use positives over negatives; experiment with synonyms for variety. For V6/V7, prompts handle longer descriptions better, with improved accuracy in depicting words and characters. Tools like /describe (Discord) can generate prompts from uploaded images.
| Prompt Element | Description | Example |
| Subject | Main focus | “golden retriever puppy” |
| Attributes | Details like color, mood | “fluffy, playful” |
| Scene/Environment | Setting | “autumn forest with falling leaves” |
| Style/Aesthetic | Artistic influence | “cinematic, in the style of Pixar” |
| Lighting | Mood enhancer | “soft golden hour sunlight, volumetric rays” |
| Technical | Camera/effects | “35mm lens, shallow depth of field, 8k resolution” |
Advanced techniques: Use “A as B” for creative twists (e.g., “city skyline as a dragon”); incorporate cinematic lighting like rim light or chiaroscuro for depth; apply camera angles (e.g., “low angle shot”) to shift perspective.
Step 4: Using Parameters for Control
Parameters are added at the prompt’s end with — (double dash), allowing fine-tuning without altering the text description. Focus on essentials first.
| Parameter | Function | Range/Example | Impact |
| –ar | Aspect ratio | –ar 16:9 | Changes image shape (e.g., widescreen for landscapes) |
| –s or –stylize | Artistic level | 0-1000 (default 100) | Higher for creative flair, lower for literal interpretations |
| –chaos | Variation diversity | 0-100 | Higher for more unpredictable grids |
| –q | Quality | 0.25-2 | Higher for details (uses more GPU time) |
| –no | Exclude elements | –no trees, blur | Attempts to remove specified items |
| –seed | Reproducibility | –seed 12345 | Same seed + prompt yields similar results |
| –cref | Character reference | –cref [image URL] –cw 80 | Ensures consistent characters in V6+ |
| –sref | Style reference | –sref [image URL] –sw 50 | Matches visual style from a reference |
| –weird | Edgy elements | 0-3000 | Adds unconventional twists |
| –iw | Image weight | 0.5-2 | Strengthens influence of reference images |
For videos, use –video to animate stills. In V6+, –cref improves consistency across generations.
Step 5: Generating, Refining, and Organizing Images
- Generation: Enter your prompt and submit. Midjourney produces a 2×2 grid of four images. Monitor progress in real time.
- Refinement: In Discord, use U1-U4 to upscale, V1-V4 for variations, or 🔄 to re-roll. On web, access upscalers, zoom out, pan, Vary Region (for targeted edits like fixing hands), and the Editor for broader changes. For videos, select an image and use animation tools to create short clips.
- Personalization: Rate images (👍/👎) to train a custom profile; use –p for personalized generations or Style Tuner for custom codes.
- Organization: Download high-res images from the Organize Page. Create folders, filter by date/parameters, or share in community rooms. Explore https://www.midjourney.com/explore for inspiration.
Advanced Features and Tips
- Image References: Upload or link images as prompts (e.g., for blending 2-5 images via /blend in Discord). Use –iw to weigh their influence.
- Video Creation: Turn images into videos by adding a starting frame or using animation parameters. V1 video model (launched June 2025) supports basic motion.
- Common Issues and Fixes: For inconsistencies like deformed hands, add “anatomically correct hands” or use q 2/Vary Region. Handle GPU limits by switching to Relax mode.
- Community and Ethics: Join Discord for tips; follow guidelines to avoid harmful content. Study successful prompts on Explore. Alternatives like DALL-E or Stable Diffusion offer free tiers but may lack Midjourney’s artistic edge.
- Updates in 2026: Watch for V8, which may introduce new UI, draft modes, and editing updates, check midjourney.com/updates regularly.
Practice iteratively: Generate, refine, and learn from variations. With time, you’ll master professional level outputs.
