midjourney（midjourney 搜索）

Midjourney Skill

Generate AI artwork via the Midjourney Discord bot. All commands are run through Discord.

⚠️ Prerequisites

1. Discord account — Required
2. Midjourney subscription — Free trials limited, paid plans at midjourney.com
3. Join Midjourney Discord — https://discord.gg/midjourney
4. Bot access — Use /imagine in #newbies channels or your own server

Quick Start

/imagine prompt: a cyberpunk cat sitting on a neon rooftop, detailed, cinematic lighting --ar 16:9 --v 6

Workflow:

1. Type /imagine in Discord
2. Enter your prompt
3. Wait ~60 seconds for 4 variations
4. Use U1-4 to upscale, V1-4 for variations

When to Use

✅ USE this skill when:

• Generating artwork, illustrations, or concept art from text
• Creating variations of an existing image
• Upscaling low-resolution images
• Blending multiple images together
• Getting prompt ideas from existing images (describe)
• Experimenting with artistic styles and parameters

❌ DON'T use this skill when:

• Need commercial licensing info → check midjourney.com/terms
• Want API access → Discord-only, no public API
• Need exact text rendering → MJ struggles with text
• Require photorealistic faces → may have artifacts

Core Commands

/imagine — Generate Images

/imagine prompt: <your description> [parameters]

Examples:

/imagine prompt: portrait of a warrior princess, golden armor, dramatic lighting, fantasy art --ar 2:3 --stylize 750

/imagine prompt: minimalist logo for a tech startup, geometric, blue and white --v 6

/imagine prompt: isometric 3D render of a cozy coffee shop, warm lighting, blender style --ar 1:1

/upscale — Increase Resolution

After generation, click U1, U2, U3, or U4 buttons under the image grid.

Or use command:

/upscale --image <job-id> --index 1

/variations — Create Variations

After generation, click V1, V2, V3, or V4 buttons.

Or use command:

/variations --image <job-id> --index 1 --strength subtle

/blend — Merge Images

/blend

Upload 2-5 images. Midjourney merges them stylistically.

Example use:

• Blend a photo with a painting style
• Merge character concepts
• Combine color palettes

/describe — Analyze Image

/describe

Upload an image → get 4 prompt suggestions that would create similar results.

Use for:

• Reverse-engineering styles
• Learning prompt patterns
• Getting inspiration from existing art

/remix — Modify Prompts

Enable remix mode to change prompts when creating variations:

/remix

Then use V1-V4 buttons with modified prompts.

Parameters Reference

Aspect Ratio (--ar)

--ar 1:1      # Square (default)
--ar 16:9     # Widescreen
--ar 9:16     # Portrait (phone)
--ar 2:3      # Classic portrait
--ar 3:2      # Landscape photo

Version (--v)

--v 6         # Latest (default)
--v 5.2       # Previous version
--v 5.1       # Older, more artistic
--niji 6      # Anime/manga style

Stylize (--s or --stylize)

--s 100       # Low stylization (follows prompt closely)
--s 250       # Default
--s 750       # High stylization (more artistic freedom)
--s 1000      # Maximum

Chaos (--chaos)

--chaos 0     # Consistent results (default)
--chaos 50    # More varied
--chaos 100   # Maximum variation

Quality (--quality or --q)

--q 0.5       # Half quality, faster
--q 1         # Default
--q 2         # Double quality, slower, more detail

Seed (--seed)

--seed 1234   # Reproducible results

Same seed + same prompt = similar images.

No (--no)

--no text     # Avoid text in image
--no blur     # Avoid blur
--no people   # Avoid people

Style Raw (--style raw)

--style raw   # Less opinionated, more literal prompt interpretation

Weird (--weird)

--weird 0     # Normal (default)
--weird 500   # Unconventional compositions
--weird 3000  # Maximum weirdness

Advanced Techniques

Multi-Prompt (::)

Weight different parts of your prompt:

/imagine prompt: cat::2 dog::1
# Cat is 2x more important than dog

Image Prompts

Use an image URL as part of your prompt:

/imagine prompt: https://example.com/image.jpg a futuristic cityscape

Character Reference (--cref)

/imagine prompt: character in different scene --cref <image-url>

Maintains character consistency across images.

Style Reference (--sref)

/imagine prompt: original content --sref <image-url>

Applies the style of a reference image.

Workflow Examples

Logo Design

/imagine prompt: minimalist tech logo, geometric shapes, gradient blue, vector style --ar 1:1 --v 6
# Review 4 options
# Click U2 to upscale best one
# Click V2 for variations of that style
# Repeat until satisfied

Character Concept Art

/imagine prompt: full body character design, fantasy rogue, leather armor, daggers, dynamic pose --ar 2:3 --niji 6
# Upscale favorite
# Use /variations for pose variations
# Use --cref to maintain character in new scenes

Product Visualization

/imagine prompt: product photography, wireless headphones on marble surface, soft studio lighting, commercial --ar 4:5 --q 2

Album Cover

/imagine prompt: psychedelic album cover, abstract swirls, vibrant colors, 1970s style --ar 1:1 --stylize 500

Tips

1. Be specific but not restrictive — "cinematic lighting" good, "exactly 47 degree angle" bad
2. Use artist names carefully — "in the style of Greg Rutkowski" works, but consider living artists ethically
3. Iterate — First result rarely perfect. Use V buttons to explore variations
4. Save job IDs — Useful for referencing later
5. Check Discord — All generations saved in your Discord DMs with bot
6. Use --v 6 — Latest version has best prompt understanding
7. Niji for anime --niji 6 for manga/anime style

Troubleshooting

"Queue full" — Wait a few minutes, try again during off-peak hours

"Invalid parameter" — Check parameter syntax, some require specific values

"Content policy violation" — Adjust prompt to avoid restricted content

"Can't upscale" — Job may have expired (24hr limit on free accounts)

"Text looks wrong" — MJ struggles with text. Try --no text or use DALL-E 3 for text-heavy images

"Faces look distorted" — Common issue. Try --v 6, add "photorealistic" or use inpainting

Pricing (as of 2026)

• Free trial — ~25 images (one-time)
• Basic — $10/mo, ~200 images
• Standard — $30/mo, unlimited relaxed mode
• Pro — $60/mo, faster, more concurrent jobs
• Mega — $120/mo, maximum capacity

Check midjourney.com for current pricing.

Resources

• Official docs: docs.midjourney.com
• Discord: discord.gg/midjourney
• Community showcase: midjourney.com/showcase
• Prompt guide: docs.midjourney.com/docs/prompt-guide