BIG Update - Midjourney Will Never Be the Same Again
TLDRMidjourney introduces a groundbreaking personalization feature that tailors image generation to individual preferences. By using a personal code, users can influence the style of generated images, enhancing the creative process. The feature is off by default but can be activated in settings, and the strength of personalization can be adjusted. The video also highlights how aspect ratios and chaos values can affect the outcome, encouraging users to experiment with these parameters for unique results.
Takeaways
- 😲 Midjourney has introduced a new personalization feature that tailors image generation to individual user preferences.
- 🔑 Users can activate personalization through a code or by adding '-dp' at the end of their prompts to see personalized results.
- 🛠️ Personalization is turned off by default and can be enabled in the user's settings within the Midjourney platform.
- 🎨 The personalization feature learns from the images a user likes on the explore page and their choices in ranking pairs.
- 🔢 It's suggested that around 200 images need to be ranked for the bot to develop a user's personal preference.
- 📈 Users can develop their own personalization code by interacting with the system and making selections based on their preferences.
- 🌟 The personalization code can significantly alter the outcome of image generation, even with short prompts.
- 🔄 The strength of personalization can be adjusted with the '-s' parameter, which has a range from 0 to 1000.
- 🔄 Experimenting with different 'chaos' values alongside personalization can yield diverse and interesting results.
- 🖼️ The aspect ratio of an image can influence how personalization affects the final generation.
- 🐶 Personalization also works with 'style raw', allowing users to incorporate their preferences into different styles of image generation.
Q & A
What is the new feature introduced by Midjourney?
-Midjourney introduced a personalization feature that allows the bot to take into account a user's personal preferences when creating a generation.
How can users personalize their prompts with Midjourney's new feature?
-Users can personalize their prompts by adding '-d-p' at the end of their prompts, which will activate the personalization code.
What is the default setting for personalization in Midjourney?
-Personalization is turned off by default but can be enabled in the user's settings.
How does Midjourney develop a personal style code for a user?
-The personal style code is developed by the model learning from the images a user likes on the explore page and the images chosen in the rank pairs.
How many images are needed for Midjourney to develop a user's personal preference?
-The Midjourney team suggests that around 200 images are needed for the bot to develop a user's personal preference.
What is the purpose of the stylized value in Midjourney's personalization feature?
-The stylized value controls the strength of the personalization, with a range between 0 and 1,000, allowing users to adjust how much their preferences influence the generation.
What is the default stylized value in Midjourney's personalization feature?
-The default stylized value is 100, which represents the base strength of personalization.
How does the aspect ratio affect the personalization in Midjourney's feature?
-The aspect ratio can change the idea of the preference quite a bit, potentially altering the generated images significantly.
Does the personalization feature work with longer prompts in Midjourney?
-Yes, the personalization feature can also have a significant impact on longer prompts, not just short ones.
How does the chaos value interact with the personalization feature in Midjourney?
-The chaos value can be experimented with when the personalization code is enabled, potentially leading to unexpected and creative results.
Is the personalization feature compatible with style raw in Midjourney?
-Yes, the personal code also works with style raw, allowing the feature to tap into the user's preferences for different styles.
Outlines
🤖 Personalization Feature Introduction
Mid Journey has introduced a personalization feature that tailors AI-generated content to individual user preferences. The bot learns from the images users like and rank on the explore page, developing a unique code for each user. This code can be applied to prompts to generate personalized images. The video demonstrates how to activate this feature in settings or by adding '-dP' to prompts. It also shows the process of developing a personal code by ranking image pairs and how this affects the generation of images based on the user's style.
🔧 Customizing Personalization Strength
The strength of personalization can be adjusted using a 'stylized' value ranging from 0 to 1,000, with 100 being the default. The video provides examples of how varying this value affects the generated images, showing that higher values intensify the personal style. It suggests that values between 50 and 200 are typically sufficient, as increases beyond this range yield diminishing returns in terms of stylistic differences. The video also touches on the potential impact of the 'chaos' value when personalization is enabled, demonstrating that higher chaos values can still produce quality images.
🖼️ Aspect Ratio and Longer Prompts Impact
The video script discusses how the aspect ratio can significantly influence the outcome of personalized image generation, with wider ratios potentially altering the 'painterly vibe' of the images. It also highlights that the personalization feature is not limited to short prompts; it can also significantly affect the results of longer, more detailed prompts. Additionally, the script mentions that personal codes can be used with 'style raw' to incorporate user preferences into different styles of image generation. The video concludes with a reminder that personal codes are not compatible with 'nii journey' and encourages users to explore and share their codes.
Mindmap
Keywords
💡Personalization feature
💡Seed number
💡Stylized value
💡Chaos value
💡Aspect ratio
💡Style raw
💡Rank pairs
💡Preference learning
💡Short prompts
💡Longer prompts
Highlights
Midjourney introduces a new personalization feature that tailors image generation to individual user preferences.
Personalization code (e.g., 'wmL1 AI') can be added to prompts to generate images that align with a user's aesthetic preferences.
Personalization is turned off by default but can be enabled in user settings.
Users can develop their personal code by ranking images they like on the explore page and in rank pairs.
The model requires around 200 image selections to learn a user's preferences effectively.
The personalization strength can be adjusted using the 'd-s' parameter, with a range from 0 to 1000.
Higher stylized values (e.g., 50-200) significantly influence the generation to match user preferences.
Experimenting with the 'chaos' value can yield unexpected and creative results even with personalization enabled.
Aspect ratio can influence how personalization affects the generated images, with wider ratios potentially altering the aesthetic.
Longer prompts can also benefit from personalization, leading to significantly different image outcomes.
Personal codes work with 'style raw' to incorporate user preferences into the image generation process.
Personal codes do not currently work with 'nii journey'.
The personalization feature is off by default, offering users the choice to opt-in for a tailored experience.
The feature is part of an ongoing evolution, with the potential for changes as the algorithm and user preferences develop.
Users are encouraged to rank pairs and share their personal codes to explore the feature's capabilities.
The video provides a comprehensive guide on how to use and experiment with Midjourney's personalization feature.