Midjourney V7 Alpha: Is the New Model Better? Early Look at Features

Midjourney's V7 Alpha model is here, bringing significant changes and new features. This version is still developing, with more updates coming soon. Let's look at the first impressions and key differences from previous versions like V6 and 6.1.

First Impressions on Image Quality and Style

One of the biggest changes people see is better image quality and detail. Many users get more beautiful and realistic images on average with V7.

The default style in V7 seems brighter and less "grungy" than V6 or 6.1. It tends to favor realism more when you don't specify a particular style in your prompt.

While realism is strong, getting consistent high quality in illustration styles can be harder right now. Expect this to improve as the model is refined.

Body Coherence and Faces

Body coherence is a bit better, though still not perfect. You might still see strange hands or distorted faces sometimes, but less often than before. Images with small faces are also slightly better.

Less Beauty Bias

V7 shows less of a "beauty bias" compared to 6.1. You'll see more natural skin textures instead of overly smoothed looks.

Prompting with V7

Midjourney V7 might require small adjustments to your prompting style. Prompts that worked well in V6 might need tweaks for V7. This is typical with new model releases.

Like earlier versions, V7 works well with clear, grammatically correct sentences. You can also use comma-separated phrases.

V7 handles longer prompts better, so feel free to include several distinct ideas in one prompt.

Prompt Adherence

Generally, V7 shows better understanding and adherence to prompts compared to 6.1. If Midjourney isn't capturing a specific detail, try describing it in a few different ways within the prompt to make it clearer.

Other Prompt Improvements

V7 is much better at handling multiple languages. It's also better at generating full body shots, even with wide images.

Speed and Relax Mode Changes

Jobs in V7 run faster than in V6, even in Relax mode. If you use the Fast mode with V7 in the settings, it will use your Turbo hours, so be mindful of that. The speed mode limit is temporary.

Midjourney V7 Personalization

To use V7, you need to set up your V7 global personalization profile. This creates a custom model based on your taste. It takes about 5 minutes to set up by picking your preferred images from at least 200 pairs on the personalize page. Once done, personalization is on by default.

V6 personalization codes can be used with V7, but V7 codes do not work with V6.

Introducing Draft Mode

Draft mode is a new feature for quick idea iteration. It lets you render images very fast at half the cost of a normal job. It's designed to keep you in a creative flow.

Draft mode has microphone capability and an AI that helps turn your ideas into prompts, even if you speak disjointed thoughts. You can even tell it to only run commands when you say a specific key phrase.

The price for this speed is lower image quality. Once you like a draft image, you can use the "Very Subtle" or "Enhance" buttons to run a higher quality version in the regular non-draft mode.

Exploring new features and generating ideas quickly with Draft Mode can be a game-changer. To take your Midjourney workflow further and automate repetitive tasks, check out TitanXT's Midjourney Automation Suite. It can help streamline your creative process.

Features and Parameters Not Available Yet in V7 Alpha

Since V7 is in alpha, some features and parameters from V6.1 are not working or available yet. Midjourney plans to add many of these back in future updates.:

• Character reference, character weight, and the upcoming Omni reference are expected soon.

• The '--no' parameter doesn't seem to work currently.

• There is no quality parameter like Q4 or Q8 yet.

• '--stop', '--tile', and '--weird' parameters do not work.

• Remix mode is not available.

• Multi-prompting for weighing parts of your prompt is not available.

• Upscaling, editor jobs, retexture, zoom, and inpaint features currently fall back to the 6.1 versions.

Style Raw (`--style raw` or `--raw`) is available.

Community Ideas and Model Development

Midjourney is community-funded and considers user feedback strongly. Features like improved text rendering were not the focus of V7 because other features were ranked higher by the community. If you want a specific feature, visit midjourney.com/ideas and participate in ranking ideas.

The Relaxathon Event Has Ended

The special Relaxathon event which gave basic plan users access to Relax mode has finished. Basic plan users will now use Fast or Turbo mode which consumes monthly GPU hours.

For plans that include Relax mode normally, batch, permutation prompts, and the `--repeat` parameter can no longer be used in Relax mode. They still work in Fast or Turbo mode.

Generating many images or running variations is often part of the creative process. If you need to handle large volumes of prompts or manage your generated images efficiently across different modes or parameters, explore what TitanXT's Midjourney Automation Suite can do to help automate these tasks.

Overall Impressions and What's Next

V7 Alpha shows exciting improvements in image quality, detail, and prompt adherence. Improvements are still needed for consistent illustration styles and complex scenes with many characters. Midjourney has updates planned to address these areas and bring back missing features.

If you're finding it challenging to adapt prompts that worked in V6 to V7, consider refining your prompt library. Tools that help manage and iterate on prompts can be incredibly useful during model transitions. Discover how TitanXT's Midjourney Automation Suite can support prompt engineering and workflow management.

If V7 isn't meeting your needs right now, you can continue using 6 or 6.1 and try V7 again after more updates are released.

Midjourney will continue to release updates frequently, so expect things to change quickly as V7 moves beyond its alpha stage.