Midjourney V7 Guide 2025 new version
Revolutionary sketch mode, voice-generated drawing, real-time editing
๐ Quick Overview
๐ Release time
April 4, 2025 (Alpha test)
๐ Core functions
Sketch mode, voice drawing, conversational interaction
โก Speed increase
Sketch mode 10x faster, half the cost
๐ Access method
Discord / midjourney.com
๐ V7 upgrades
โ๏ธDraft Mode Revolutionary
The biggest highlight of V7, changing the way of AI painting interaction
- Speed increase: Generates 10 times faster than standard mode
- Half the cost: Resource consumption is only 50% of standard mode
- Conversational interaction: Modify pictures like chat
- Real-time editing: "Change the cat into an owl", "Change to a night scene"
๐ค Voice Picture
Just speak and the picture will be generated!
- Click the microphone icon to start
- Support Chinese speech recognition
- Adjust while talking, shaping the image in real time
๐ค Personalized model (enabled by default)
AI learns your aesthetic preferences
- Completed 200 sets of picture preference tests
- Create an exclusive "visual profile"
- Automatically reference your style every time it is generated
โ๏ธ Sketch Mode Usage Guide
Switch to V7
Enter in Discord
/settings
, select V7 model
Enable sketch mode
The web editor will automatically switch to a "conversational interactive interface"
Generate first draft
Enter prompt words to quickly generate low-resolution sketches
Conversational modification
Use natural language to describe modification requirements, such as "make the background darker"
Upgrade to HD
When satisfied, use the "Enhance" function to generate a high-resolution version
๐ Omni-Reference Omni-directional reference
V7โs new powerful image reference system surpasses V6โs character reference:
๐ฏ Precise element specification
Upload the reference image and specify "Put this element into the image"
๐ญ Diversified support
Characters, animals, props, vehicles, costumes, styles
๐ฅ Multi-object generation
A single image contains multiple reference objects
โ๏ธ Weight adjustment
--ow parameter controls reference influence strength
Weight parameter example
| --ow 25 | Low weight, suitable for style conversion (photo โ animation) |
| --ow 100 | Default value, balance reference and creativity |
| --ow 400 | High weight to ensure high consistency of face/clothing |
| --ow 1000 | Maximum weight, almost exact copy of reference |
โฌ๏ธ V7 regular upgrade
๐ Text understanding enhancement
Understand complex prompt words more accurately
๐จ Detailed texture upgrade
More refined material and texture performance
โ Hand/limb optimization
Significantly improve the character's hand structure
๐ Object continuity
The relationship between objects is more natural
๐ V6 vs V7 comparison
| Features | V6 | V7 |
|---|---|---|
| Interaction mode | Traditional Prompt | Conversational + Voice |
| Iteration speed | Standard | Sketch mode 10x faster |
| Image Reference | role reference (cref) | Omni Reference (Omni) |
| Personalization | Manual setting | Enable learning by default |
| Hand performance | Occasional problems | Significant improvement |
| Real-time editing | โ | โ |
| Voice input | โ | โ Support Chinese |
โ๏ธ V7 common parameters
| --draft | Enable sketch mode, support multi-version iteration and batch generation |
| --ow [0-1000] | Omni-Reference weight, controls the reference influence strength |
| --stylize [0-1000] | Degree of stylization, the higher the value, the more artistic it is. |
| --sref [URL] | Style reference, maintain a certain visual style |
| --cref [URL] | Character reference to maintain character consistency |
| --ar [ratio] | Screen ratio, such as 16:9, 9:16, 1:1 |
| --no [element] | exclude certain elements, such as --no text |
| --chaos [0-100] | Degree of change, the higher the value, the more diverse the results. |
๐ V7 Prompt Example
Quick iteration in sketch mode
Omni-Reference use
Voice Picture (Chinese)
Advanced prompt word template
๐ฐ Midjourney Pricing
Basic
- ~200 pictures/month
- 3 concurrent tasks
- Available for commercial use
Standard
- 15 hours rapid generation
- Infinite relaxation mode
- Available for commercial use
Pro
- Rapid generation in 30 hours
- Stealth mode
- 12 concurrent tasks
Mega
- 60 hours rapid generation
- 12 concurrent tasks
- Team/Enterprise Level
โ FAQ
When will V7 be officially released?
V7 started Alpha testing on April 4, 2025, and Midjourney said it will continue to update new features in the next 60 days (every 1-2 weeks).
What should I do if the image generated in sketch mode has low resolution?
Sketch mode is designed for rapid iteration. Once you've decided on a direction you're happy with, use the "Upscale" function to upgrade to a high-definition version.
Does the voice-generated image support Chinese?
Support! The speech recognition function of V7 already supports Chinese, and you can directly describe the image you want to generate in Chinese.
How to train a personalized model?
The first time you use V7, you need to complete about 200 sets of picture preference tests (select the pictures you like more), and the system will automatically learn your aesthetic preferences.
Is V7 more expensive than V6?
Subscription price remains unchanged. But the sketch mode consumes less money, which means it can generate more pictures for the same price.
How to switch back to V6?
You can switch between V6 and V7 using the /settings command in Discord. Or add --v 6 to the end of the prompt word.