Midjourney v6.1 vs v6.0: What You NEED to Know | SREF, Beauty Bias, New --p Codes | Midjourney Guide

Woollyfern
8 Aug 202406:18

TLDRThe video compares Midjourney's v6.1 and v6.0 models, highlighting improvements in image quality and texture in v6.1, but noting a stronger conventional beauty bias. It discusses the Q2 mode for more details, upscalers, and the impact on character references. Viewers learn about the new model personalization system, improved text rendering, and seamless pattern creation. The host shares personal preferences and invites feedback on v6.1, teasing upcoming SRF codes and a Patreon prompt collection.

Takeaways

  • 🔧 Midjourney has released a new v6.1 model, which is now the default for all users, but can be changed in settings.
  • 🖼️ The v6.1 update enhances image quality with reduced pixel artifacts and improved textures, especially in eyes and skin.
  • 🔍 V6.1 introduces a Q2 mode for more detailed textures at the expense of slight image coherence reduction.
  • 🆕 New 2X upscalers are included in v6.1, offering a different approach to image enhancement compared to Topaz Labs' Gigapixel.
  • 🎨 V6.1 has a more pronounced conventional beauty bias, resulting in smoother skin and fewer imperfections, which Midjourney is aware of and plans to address in v7.
  • 👤 Users can still create images of average or unique-looking people with v6.1, but it might be easier to stick with v6.0 for now.
  • 🤖 Character references work similarly in v6.1 as in v6.0, with no significant changes in how they are handled.
  • 🎭 Style references (SR codes) maintain the same style between v6.0 and v6.1, but v6.1 images may appear less detailed.
  • 📝 Text rendering accuracy is improved in v6.1, though the speaker admits to limited personal experience with this feature.
  • 🧩 The 'tile' parameter for creating seamless patterns works better in v6.1, especially when upscaling images.
  • 🤖 The model personalization system in v6.1 is more nuanced and accurate, with codes that change as users rank more images.

Q & A

  • What is the main focus of the video script provided?

    -The video script focuses on comparing the new Midjourney v6.1 model with its predecessor, v6.0, highlighting the differences in image quality, conventional beauty bias, and new features like the --p code and model personalization.

  • How does the image quality differ between Midjourney v6.1 and v6.0?

    -The v6.1 update enhances image quality with reduced pixel artifacts, improved textures, clearer eyes and skin, and better coherence and accuracy in elements like bodies and hands.

  • What is the purpose of the --Q2 parameter in Midjourney v6.1?

    -The --Q2 parameter in Midjourney v6.1 is used to increase texture and detail in the results, albeit at the cost of slightly reduced image coherence.

  • How does Midjourney v6.1 compare to Topaz Labs' Gigapixel in terms of upscaling?

    -The video script does not definitively state which is better, but it suggests that Midjourney's upscaler and Topaz Labs' Gigapixel both perform well, with Topaz showing a slight edge in skin texture and eye clarity in the provided comparison.

  • What is the conventional beauty bias mentioned in the script, and how does it affect Midjourney v6.1?

    -The conventional beauty bias refers to the tendency of v6.1 to produce images with smoother skin and fewer imperfections. This was an unintended consequence of incorporating user-generated data to improve image quality, and the developers are working to correct it in the upcoming v7.

  • Can character references from previous versions be used with Midjourney v6.1?

    -Yes, character references from previous versions can be used with v6.1, and the model handles character references similarly to v6.0 without any updates to the character reference model.

  • How does the style rendering with style references (Sr codes) differ between Midjourney v6.0 and v6.1?

    -While the style remains consistent between v6.0 and v6.1, v6.1 images tend to be more simplistic and less detailed compared to v6.0.

  • What improvements are expected in the text rendering accuracy in Midjourney v6.1?

    -Midjourney v6.1 is expected to have improved text rendering accuracy, handling longer phrases with fewer mistakes and more complicated phrases more effectively.

  • How does the new model personalization system in Midjourney v6.1 work?

    -The new model personalization system in v6.1 is more accurate and nuanced, interpreting visual aesthetic preferences based on image rankings. Users can see their personalized codes by typing /list personalized codes in Discord, and the codes change as more images are ranked.

  • Can old and new personalization codes be combined in Midjourney v6.1?

    -Personalization codes can be combined, but they must all be from the same system. You cannot combine an old personalization code from v6.0 with a new one from v6.1.

  • What is the speaker's personal preference regarding the aesthetics of Midjourney v6.0 and v6.1?

    -The speaker personally prefers the aesthetic of v6.0 for its creativity, detail, and diversity, but acknowledges the improved quality and clarity of v6.1 and is looking forward to the release of v7.

Outlines

00:00

🖼️ Mid Journey v6.1 Model Review

This paragraph reviews the new v6.1 model of Mid Journey, comparing it with the previous 6.0 version. The v6.1 model is now the default for all users, but can be changed in settings. The update enhances image quality, reducing pixel artifacts and improving textures, especially in eyes and skin for more realistic photographic images. It introduces a Q2 mode for additional texture and detail at the cost of reduced coherence. A comparison with Topaz Labs' Gigapixel upscaler shows varying results, with Mid Journey excelling in certain areas but Topaz providing better overall skin texture and clarity. The paragraph also mentions a conventional beauty bias in v6.1, which Mid Journey is aware of and plans to address in v7. Additionally, character references work similarly in both versions, and style references (Sr codes) maintain the same style but with less detail in v6.1.

05:00

🔍 Exploring Mid Journey v6.1 Features and Personalization

The second paragraph delves deeper into the features of Mid Journey's v6.1 model. It discusses improvements in text rendering accuracy and the seamless pattern creation with the new tile parameter. The paragraph also introduces the new model personalization system, which is designed to be more accurate and nuanced in interpreting visual aesthetic preferences based on user image rankings. The personalization code evolves as more images are ranked, and users can view all their personalized codes in Discord. The paragraph concludes with the presenter's personal preference for the 6.0 aesthetic due to its creativity, detail, and diversity, and expresses anticipation for the upcoming v7 release. The presenter also mentions upcoming SRF codes and a monthly prompt collection for Patreon members, inviting viewers to subscribe and engage with the content.

Mindmap

Keywords

Midjourney v6.1

Midjourney v6.1 refers to the updated version of the Midjourney AI model, which is a tool used for generating images based on textual prompts. The video discusses the improvements and differences between this version and its predecessor, v6.0. It is central to the video's theme as it sets the stage for the comparison and highlights the advancements in image quality and features.

Image Quality

Image quality is a critical aspect of the video, focusing on the enhancements made in Midjourney v6.1. It includes reduced pixel artifacts, more realistic textures, and improved clarity in eyes and skin. The script mentions that these improvements result in more photographic and coherent images, which is vital for users seeking high-quality visual outputs.

Pixel Artifacts

Pixel artifacts are unintended visual distortions that can occur in digital images, often appearing as blocky or blurry regions. The script notes that Midjourney v6.1 has reduced these artifacts, leading to cleaner and more natural-looking images, which is a significant improvement over v6.0.

Q2 Mode

Q2 mode is a quality parameter in Midjourney that allows for more detailed and textured images at the potential cost of reduced coherence. The script explains that by adding '--Q2' to a prompt, users can achieve richer visual details, illustrating the flexibility of the AI in catering to different artistic preferences.

Upscaler

An upscaler is a tool that increases the resolution of an image while attempting to maintain or improve its quality. The video compares Midjourney's 2X upscaler to Topaz Labs' Gigapixel, noting differences in how they handle image details like eyelashes and skin texture. This comparison is relevant as it provides insights into the capabilities of the AI's image enhancement features.

Conventional Beauty Bias

Conventional beauty bias refers to the tendency of an AI to produce images that adhere to widely accepted standards of attractiveness, often resulting in smoother skin and fewer imperfections. The script points out that v6.1 has a more pronounced conventional beauty bias than v6.0, which could be a limitation for those seeking more diverse or unique character representations.

Model Personalization

Model personalization is a feature in Midjourney that allows users to train the AI to better understand and replicate their visual aesthetic preferences based on image rankings. The script discusses the improvements in accuracy and nuance in v6.1, highlighting the system's adaptability to individual user tastes.

SR Codes

SR codes, or style reference codes, are specific parameters used in Midjourney to guide the AI in creating images in a particular artistic style. The video script explores how these codes perform in v6.1, noting that while the style remains consistent, there may be less detail and complexity compared to v6.0.

Text Rendering

Text rendering accuracy refers to how well the AI can interpret and visually represent text within an image. The script mentions that v6.1 has improved text rendering, which is important for users who incorporate text into their image prompts and seek accuracy in longer or more complex phrases.

Seamless Pattern

A seamless pattern is a repeating design where the pattern tiles fit perfectly together without visible breaks. The script notes that v6.1 has fixed issues with upscaling images that disrupt seamless patterns, allowing for higher-resolution tiles without breaking the continuity of the design.

Personalization Code

Personalization code is a unique identifier in Midjourney's model personalization system that reflects an individual user's aesthetic preferences. The script explains that these codes change as users rank more images and can be used in both v6.0 and v6.1, providing users with the ability to experiment with different personalization codes to achieve desired results.

Highlights

Midjourney's v6.1 model is now the default for all users, but can be changed in settings.

v6.1 enhances image quality with reduced pixel artifacts and improved textures.

Eyes and skin in v6.1 look more realistic and have better clarity in photographic images.

v6.1 offers better coherence and accuracy for bodies and hands in images.

The --Q2 mode in v6.1 can provide more texture and details at the cost of slightly reduced image coherence.

New 2X upscalers in v6.1 are compared to Topaz Labs' Gigapixel for image quality.

v6.1 has a more conventional Beauty bias, resulting in smoother skin and fewer imperfections.

User-generated data was used to improve v6.1's image quality, but it also increased the Beauty bias.

v6.1's Beauty bias is acknowledged and will be addressed in the upcoming v7.

For unique looking characters, v6.0 might be preferred over v6.1 due to less Beauty bias.

Character references work similarly in both v6.0 and v6.1.

SR codes maintain the same style in v6.1 but with less detail and complexity.

Text rendering accuracy is improved in v6.1, especially for longer and more complicated phrases.

The tile parameter in v6.1 creates seamless patterns that do not break upon upscaling.

The new model personalization system in v6.1 is more accurate and nuanced.

Personalization codes change as you rank more images in v6.1 and can be viewed with /list personalized codes.

Old and new personalization codes can be used in both v6.0 and v6.1, but not combined across versions.

The presenter prefers v6.0 for its creativity, detail, and diversity, anticipating improvements in v7.

A new set of SRF codes and a monthly prompt collection are upcoming for Patreon members.

Casual Browsing

Stable diffusion VS Midjourney: All you need to know

2024-05-18 10:20:02

Midjourney Copyright Law: What AI Artists Need to Know

2024-05-03 17:05:01

6 Essential Midjourney Parameters You Need To Know

2024-05-21 02:40:01

Midjourney On The Web - All You Need To Know | Midjourney Alpha

2024-06-13 02:45:01

The Midjourney Showcase in-depth: All you need to know.

2024-05-16 20:30:03