BIG Update - Midjourney Will Never Be the Same Again

Summary

Mid Journey introduces a new personalization feature that tailors image generation to individual preferences. Users can activate this feature through settings or by adding a personal code to their prompts. The system learns from the images users like and rank, developing a unique style code. This code can be tested and adjusted with a stylized value to control the strength of personalization. The video demonstrates the impact of personalization on image generation, showcasing how it can enhance short and long prompts and work with different styles, including style raw. The aspect ratio and chaos value are also shown to influence the outcome, with the video encouraging experimentation and reminding viewers that the feature is evolving.

Takeaways

• 😀 Mid Journey has introduced a personalization feature that tailors image generation to individual user preferences.
• 🔄 To utilize personalization, users can add a personal code (e.g., '-wml1 AI') to their prompts or enable it in settings.
• 🛠️ Personalization is turned off by default but can be activated in the user's settings to apply the personal code to all prompts.
• 🎨 The personal code is developed by the system based on the user's preferences in image selection on the explore page and ranking pairs.
• 📈 Users need to rank around 200 images for the bot to learn their preferences effectively.
• 📝 The script provides a step-by-step guide on how to develop and test a personal code for image generation.
• 🌟 The personalization feature significantly impacts both short and long prompts, altering the generated images to match user preferences.
• 🔧 Users can control the strength of personalization using a 'stylized' value ranging from 0 to 1000, with 100 being the default.
• 🔄 Experimenting with different 'chaos' values in conjunction with personal codes can yield diverse and unexpected image results.
• 📐 The aspect ratio of the image can influence how the personalization feature interprets and applies user preferences.
• 🐶 Personal codes also work with 'style raw' prompts, adjusting the generated images to align with user-designated preferences.

Q & A

What is the new feature introduced by Mid Journey?

The new feature introduced by Mid Journey is personalization, which allows the bot to create generations based on the user's personal preferences.

How does the personalization feature work when generating images?

The personalization feature works by appending a personalization code (e.g., '-d-p') to the end of the user's prompt, which then generates images tailored to the user's preferences.

What is the default setting for the personalization feature?

The personalization feature is turned off by default but can be enabled in the user's settings.

How can users develop their own personalization code?

Users can develop their own personalization code by interacting with the model on the Mid Journey website, ranking images they like on the explore page, and choosing images in the rank pairs. Around 200 images are needed for the bot to develop the user's personal preference.

How can users adjust the strength of the personalization in their images?

Users can adjust the strength of the personalization by using the '-d-s' parameter followed by a value between 0 and 1000, where 100 is the default.

What is the role of the 'chaos' value in the personalization feature?

The 'chaos' value, when enabled with the personalization code, can introduce unexpected details and variations in the generated images, enhancing the creativity of the results.

How does the aspect ratio affect the personalization feature?

The aspect ratio can significantly change the outcome of the personalization feature, as it may alter the interpretation of the user's preferences in the generated images.

Can the personalization code be used with the 'style raw' command?

Yes, the personalization code can be used with the 'style raw' command, allowing the feature to tap into the user's preferences even in different style applications.

Is there a limit to how high the 'chaos' value can go?

While there is no explicit limit, the script suggests that going beyond a certain point (e.g., 200) may not significantly change the results, indicating diminishing returns.

How does the personalization feature handle longer prompts?

The personalization feature can also significantly impact longer prompts, altering the generated images to better match the user's preferences even when more detailed descriptions are provided.

Why might the personalization feature not work with 'nii journey'?

The script does not provide a specific reason, but it mentions that personal codes do not work with 'nii journey' at this time, suggesting that compatibility or feature implementation may be limited.