Midjourney V6 is Too Good Now... (ALL New Updates Explained)

AI Catalyst
8 Feb 202404:22

TLDRMidjourney V6 introduces a plethora of updates that enhance performance and usability. The model is now faster and more cost-effective, with upscalers operating at double the speed and half the cost. Image aesthetics, coherence, and text rendering have improved significantly. New tools like pan, zoom, and VAR region are available for image manipulation. A new 'nii' model caters to Eastern and Anime aesthetics. The standout feature is the style reference, allowing users to upload an image as a style guide for new creations. The website has also seen improvements, offering a more user-friendly experience. Despite being in alpha, Midjourney V6 promises even more features in the future.

Takeaways

  • ๐Ÿš€ Midjourney V6 has introduced significant updates since its December release, enhancing speed, cost-efficiency, and image quality.
  • ๐Ÿ” The upscaler in V6 is now twice as fast and costs half as much, reducing GPU hour consumption.
  • ๐ŸŽจ Image aesthetics, coherence, prompt accuracy, and quality have been notably improved from the initial release.
  • ๐Ÿ“ Text rendering on images has been enhanced, allowing for better word generation within images, but still requires specific prompt formatting.
  • ๐Ÿ› ๏ธ V6 supports previous version tools like pan, zoom, and VAR region, offering more editing capabilities for generated images.
  • ๐Ÿ”„ The pan tool allows for image expansion in a chosen direction, while the zoom tool offers the ability to adjust the image's scale.
  • ๐Ÿ“ The VAR region tool enables specific image areas to be selected and transformed, though prompt adherence is not perfect.
  • ๐ŸŒ A new 'nii' model has been released, tailored for Eastern and Anime aesthetics with a higher stylized value.
  • ๐Ÿ–ผ๏ธ The style reference feature allows users to upload an image to Discord and use it as a style reference for generating new images.
  • ๐Ÿ”‘ The 'd-s ref' command is used in prompts to incorporate the style of an uploaded image, with options for multiple references and stylization strength.
  • ๐ŸŒ The mid Journey website has been revamped, offering an Alpha version with advanced features for users who have generated over 5,000 images.
  • ๐Ÿ”ฎ Despite these updates, Midjourney V6 remains in the alpha stage, with more features and improvements expected in the future.

Q & A

  • What are some of the key updates introduced in Midjourney V6?

    -Midjourney V6 has introduced faster and cheaper upscalers, improved image aesthetics, coherence, prompt accuracy, and image quality. Text rendering has also been improved, allowing better word generation on images.

  • How has the performance of the upscalers changed in V6?

    -Upscalers are now twice as fast and cost half as much, consuming fewer fast GPU hours.

  • What improvements have been made to text rendering in Midjourney V6?

    -Text rendering has been noticeably improved, allowing for better word generation on images, though it still cannot generate full-blown sentences.

  • Which tools from previous versions are now supported in V6?

    -V6 now supports tools like pan, zoom, and vary region, which were not available on the release date.

  • What functionalities do the pan and zoom tools provide?

    -The pan tool allows users to expand the image in the chosen direction, while the zoom tool zooms out the image with the possibility to zoom out every next image.

  • What is the 'vary region' tool and how does it work?

    -The 'vary region' tool allows users to select a specific part of the image and transform it. However, prompt adherence with these tools is still not very great.

  • What is the new Nii model in Midjourney V6?

    -The Nii model is specifically tuned for Eastern and Anime aesthetics, with a much higher stylized value than the default version. Users can use the 'style raw' parameter for better prompt adherence.

  • How does the new style reference feature work in V6?

    -Users can upload an image to Discord and use it as a style reference in their prompt. By writing the new 'd-s ref' command and pasting the image link, the generated images will reflect the style of the reference image.

  • Can multiple images be used as style references simultaneously?

    -Yes, users can use multiple images as style references, add different weights to them, and select the overall stylization strength using the 'ssw' command.

  • What are some enhancements made to the Midjourney website?

    -The Midjourney website has been significantly improved, offering a special Alpha version with an image generation feature and full access to all tools for users who have generated more than 5,000 images.

Outlines

00:00

๐Ÿš€ Mid Journey V6 Updates Overview

The script introduces a range of updates to the Mid Journey V6 model since its December release. Enhancements include faster and more cost-effective upscaling, improved image aesthetics, coherence, and quality. Text rendering on images has also been refined, allowing for better visual word generation within prompts. The script highlights the inclusion of previous version tools such as pan, zoom, and VAR region, which were not initially available. These tools enable users to expand, zoom, and adjust the aspect ratio of images, although prompt adherence is noted to be less than perfect.

๐ŸŽจ New 'nii' Model for Stylized Aesthetics

A new 'nii' model has been introduced, specifically tailored for Eastern and Anime aesthetics. This model has a higher stylized value, offering better prompt adherence when using the style raw parameter. The script provides examples of images created using the 'nii' model, showcasing its ability to generate highly stylized artwork.

๐Ÿ“ธ Style Reference Feature Introduction

One of the most significant updates is the introduction of the style reference feature, which allows users to upload an image to Discord and use it as a style reference in their prompts. This feature enables the Mid Journey V6 to generate new images that incorporate the style of the uploaded image along with the user's prompt. The script explains how to use this feature, including adding weights to multiple style references and adjusting the overall stylization strength with the '--ssw' command.

๐Ÿ› ๏ธ Consistent Style and Upcoming Character Features

The script discusses the 'consistent style' feature and mentions an upcoming 'consistent character' feature, which is expected to function similarly to the style reference feature. It suggests that these updates are part of the ongoing development of the Mid Journey V6, which is still in the alpha stage, indicating that more features and improvements are on the horizon.

๐ŸŒ Mid Journey Website Enhancements

The script concludes by highlighting the improvements made to the Mid Journey website. Users who have generated over 5,000 images can now access an Alpha version of the site, which includes an image generation feature and full access to all tools. The website is described as being visually appealing and more convenient than the Discord bot, providing a better user experience.

Mindmap

Keywords

Midjourney V6

Midjourney V6 refers to the sixth version of an AI tool designed for image generation and enhancement. It is the central subject of the video, with updates that have significantly improved its capabilities. The video discusses how this version has become faster and more cost-effective, with enhanced image aesthetics and coherence in prompts.

Updates

In the context of the video, 'updates' refers to the new features and improvements added to Midjourney V6 since its release. These updates are crucial as they demonstrate the tool's evolving capabilities, such as increased speed and reduced costs for image upscaling.

Upscale

Upscaling in the video script pertains to the process of improving the resolution of an image without losing quality. Midjourney V6's upscaling tools are now twice as fast and consume fewer GPU hours, indicating an improvement in performance.

Image Aesthetics

Image aesthetics in this video refers to the visual appeal and artistic quality of the images generated by Midjourney V6. The script mentions that the aesthetics, along with coherence and prompt accuracy, have been noticeably improved in the V6 model.

Coherence

Coherence in the script relates to the consistency and logical connection between the elements of an image or between the image and its associated prompt. The video notes that Midjourney V6 has improved in maintaining coherence in the generated images.

Text Rendering

Text rendering is the process of generating text within an image, as discussed in the video. Midjourney V6 has shown improvements in this area, allowing for better word generation on images, although it still cannot generate full sentences.

Pan Zoom

Pan and Zoom are tools available in Midjourney V6 that allow users to manipulate the image's view. 'Pan' lets users expand the image in a chosen direction, while 'Zoom' adjusts the image's scale, with the ability to set aspect ratios for a consistent look across a series of images.

Vary Region

Vary Region is a feature that enables users to select a specific part of an image and apply transformations to it. This tool is part of the updates in Midjourney V6 and is used to edit or complement the initial prompt for generating images.

Nii Model

The Nii model mentioned in the script is a version of Midjourney specifically tuned for Eastern and Anime aesthetics. It has a higher stylized value, which means it can generate images with more pronounced artistic styles, as demonstrated by the images created by this model.

Style Reference

The style reference feature in Midjourney V6 allows users to upload an image to Discord and use it as a style reference for generating new images. This feature integrates the style of the uploaded image with the user's prompt to create a cohesive artistic output.

Consistent Style

Consistent style is a feature that allows the AI to maintain a uniform artistic style across a series of images. The video mentions that this feature is part of the updates in Midjourney V6, and it can be adjusted using the '--ssw' command to control the overall stylization strength.

Discord

Discord, in the context of the video, is a platform where the new style reference feature of Midjourney V6 is implemented. Users can upload images to Discord and then use those images as style references in their prompts for image generation.

Alpha Version

The term 'Alpha version' in the script refers to a pre-release version of the Midjourney website that offers enhanced features and tools. Users who have generated more than 5,000 images with Midjourney can access this special version, which includes full access to all tools and an image generation feature.

Highlights

Midjourney V6 has become faster and cheaper, with all upscalers now twice as fast and costing half as much.

Image aesthetics, coherence, prompt accuracy, and image quality have been noticeably improved since the release.

Text rendering on images has been enhanced, allowing for better word generation within images.

V6 now supports tools from previous versions, such as pan, zoom, and VAR region, which were not available on the release date.

The pan tool allows for expanding the image in a chosen direction.

The zoom tool enables zooming out of the image with the possibility to adjust for every next image.

The aspect ratio can be changed using the zoom value and the --AR command.

Vary region allows for selecting a specific part of the image and transforming it.

A new nii model has been released, specifically tuned for Eastern and Anime aesthetics.

The nii model has a higher stylized value than the default version, improving prompt adherence.

V6 introduces a brand new style reference feature, allowing users to upload an image as a style reference for image generation.

The style reference feature uses the style of the uploaded image and the user's prompt to generate a new image.

Multiple images can be used as a style reference, with different weights and overall stylization strength.

The consistent style feature is expected to be followed by the introduction of consistent character features.

The mid Journey website has been significantly improved and enhanced, offering a special Alpha version for users who have generated over 5,000 images.

The website offers an image generation feature and full access to all tools, providing a more convenient experience than the Discord bot.

Despite the updates, mid Journey V6 is still in the alpha stage with more features to explore in the future.