Midjourney Update | September 2024: New Editor?, Niji Personalization, Storytelling, Video & More

Woollyfern
1 Oct 202404:58

TLDRMidjourney's September update introduces significant enhancements to its image editor, allowing editing of external images and potential advanced features like converting photos to drawings. Personalization is now easier with a new page guiding users through the process. The Niji model gains personalization options, requiring users to rank 200 image pairs. Development of V7 is prioritized, promising improvements in prompt accuracy and knowledge understanding, with a release expected by year-end. The team also works on 3D and video capabilities, aiming for simple, high-quality, and affordable short animations. Additionally, a new storytelling tool is in the works, focusing on world-building with a different UI.

Takeaways

  • 🎨 Midjourney's image editor is receiving a significant update, allowing users to edit external images, not just those created within Midjourney.
  • 🛠️ There are plans to implement a controlnet feature in the image editor, potentially enabling transformations like converting photos to drawings or adjusting scene lighting.
  • 💡 Midjourney has introduced a new personalization page to help users start with personalizing their visual style preferences.
  • 📈 Users can rank image pairs to teach Midjourney their preferences, which then gets applied to the results.
  • 🌐 Midjourney has launched a community-driven image curation task where users can contribute to the community explore page.
  • 🎭 The Niji model, trained for anime-style images, now supports personalization, requiring users to rank at least 200 image pairs.
  • 🚀 Development of Midjourney version 7 (V7) is prioritized over a version 6.2 release, with an expected release by the end of the year.
  • 🧠 V7 is anticipated to improve prompt accuracy, text rendering, and understanding of both abstract and literal interpretations.
  • 🎞️ Midjourney is focusing on 3D and video capabilities, with a simple, good-quality video model expected to be released before V7.
  • 📚 A new storytelling tool is in the works, focusing on world-building rather than just image creation, with a different user interface.
  • ⏳ Starting October 1st, any purchased fast hours will expire after 60 days.
  • 📊 Midjourney is tweaking the explore feed algorithms to improve visibility of various aspect ratios and might replace 'random' and 'hot' sections with 'new' and 'top'.

Q & A

  • What is the significant update coming to Midjourney's image editor?

    -Midjourney's image editor is about to receive a big update that will allow users to edit external images, which was not possible with the initial release.

  • Are there any potential restrictions or limitations to the new image editor features?

    -Yes, there will probably be heavy moderation around editing external images, and some advanced editor features might be gated to higher tier subscriptions or those with annual plans.

  • What new feature might be implemented in the image editor that could change a photographic image into a drawing?

    -The new feature being discussed is called 'controlnet', which could allow for capabilities such as changing a photographic image into a drawing or adjusting the lighting in a scene.

  • How can users get started with personalization on Midjourney?

    -Users can get started with personalization by visiting the new personalization page on the Midjourney website, where they can teach the system their preferred visual style and preferences by ranking image pairs.

  • What is the benefit of setting up personalization for the Niji model?

    -Setting up personalization for the Niji model allows users to create a personalized anime-style model that reflects their aesthetic preferences.

  • Why has the development of V7 been prioritized over a V6.2 release?

    -V7 is in the development and training phase, and although the team hit a few challenges and bugs, they decided to prioritize V7 as it is expected to offer improvements in prompt accuracy, text rendering, and knowledge understanding.

  • What new capabilities is V7 being built with in mind?

    -V7 is being built with 3D and video capabilities in mind, with the video models showing significant progress each week.

  • What can users expect from the upcoming video model from Midjourney?

    -Users can expect short image-to-video animations from the upcoming video model, which aims to be simple, good quality, and not overly expensive.

  • What is the basic workflow envisioned for the 3D feature in Midjourney?

    -The basic workflow for the 3D feature involves creating a Midjourney image and then clicking a 'make 3D' button to move the camera around within certain limits.

  • What new tool might users get a sneak peek of in the coming weeks?

    -Users might get a sneak peek of a new storytelling tool that focuses on world-building rather than just image creation and will have a different user interface.

  • What change is being implemented regarding the expiration of fast hours purchased on Midjourney starting October 1st?

    -Starting October 1st, any fast hours purchased on Midjourney will expire after 60 days.

Outlines

00:00

🖼️ Mid Journey Updates Overview

This video provides a rundown of the latest updates for Mid Journey in September. The image editor is set to receive a significant update, allowing users to edit external images, with potential advanced features being exclusive to higher-tier subscriptions. There's also a new personalization page to help users customize their visual preferences. The video discusses the launch of a new personalization feature for the 'nii' model, which is tailored for anime-style images. Furthermore, the development of V7 is underway, with a focus on improving prompt accuracy and text rendering, and the potential for 3D and video capabilities. Lastly, there are miscellaneous updates including a new storytelling tool and changes to the community explore page.

Mindmap

Keywords

💡Midjourney

Midjourney refers to a company or platform that specializes in AI-generated content, particularly in the realm of image creation. In the context of the video, Midjourney is updating its image editor, which is a significant development for users who create and edit images using AI. The platform's updates are aimed at enhancing user experience and expanding creative possibilities.

💡Image Editor

The Image Editor is a tool provided by Midjourney that allows users to manipulate AI-generated images. The script mentions an upcoming update that will enable editing of external images, which is a significant expansion of its current capabilities. This feature is expected to be popular among users who want more control over their images.

💡Controlnet

Controlnet is a feature that is being considered for integration into Midjourney's Image Editor. It is hinted to offer advanced image manipulation capabilities, such as transforming photographs into drawings or adjusting lighting in a scene. This suggests a move towards more sophisticated image editing within the AI platform.

💡Personalization

Personalization in the context of Midjourney is a feature that tailors the AI's output to an individual's aesthetic preferences. The platform has introduced a new personalization page to facilitate this process. Users can 'teach' Midjourney their preferred visual style by ranking image pairs, which the AI then uses to generate more personalized results.

💡Niji

Niji is Midjourney's anime-trained model, designed to generate images in the style of anime or manga. The script mentions the ability to set up personalization for the Niji model, similar to the main Midjourney model, allowing users to customize their anime-style image generation.

💡V7

V7 is an upcoming version of Midjourney's AI model, which is currently in development. The script indicates that the development of V7 is being prioritized over a V6.2 release. V7 is expected to improve various aspects of image generation, including prompt accuracy and knowledge understanding, with a focus on 3D and video capabilities.

💡3D

The term '3D' in the script refers to Midjourney's plans to expand into three-dimensional content creation. The platform is wrapping up its data collection process and is about to start larger training runs for 3D models. This suggests that users may soon be able to generate 3D versions of their AI-created images.

💡Video

The script discusses Midjourney's plans to venture into video creation, with the development of video models that are showing promising results. The initial focus is expected to be on short image-to-video animations, aiming for simplicity, quality, and affordability.

💡Storytelling

The term 'Storytelling' in the script refers to a new tool that Midjourney is developing, which is expected to focus on world-building rather than just image creation. This tool will have a different user interface and is part of the platform's expansion into narrative and creative content generation.

💡Fast Hours

Fast Hours are a feature within Midjourney that allow users to expedite the processing of their image generation requests. The script mentions a change to the policy where any purchased Fast Hours will now expire after 60 days, which is a significant update to how users manage their usage of the platform.

💡Explor Feed

The Explor Feed is a feature within Midjourney that displays a variety of user-generated images. The script indicates that the platform is tweaking the algorithms to show a more diverse range of aspect ratios and potentially removing the 'random' and 'hot' sections in favor of 'new' and 'top' sections.

Highlights

Midjourney's image editor is about to receive a significant update, allowing editing of external images.

The new editor will feature heavy moderation for editing external images.

A controlnet feature may be implemented for advanced image transformations.

Advanced editor features might be restricted to higher-tier subscriptions.

Midjourney launched a new personalization page to ease the setup process.

Personalization allows Midjourney to learn and apply user preferences to image results.

Users are encouraged to rank image pairs to teach Midjourney their preferences.

Personalization setup takes approximately 5-10 minutes to generate the first code.

Ranking image pairs can earn users fast hours to speed up processing.

Personalization for the Niji model allows customization of the anime-style image generator.

V7 development is prioritized over a v6.2 release, with expected improvements in prompt accuracy and text rendering.

V7 is being developed with 3D and video capabilities in mind.

A video model may be released before V7, focusing on short image-to-video animations.

The 3D workflow will allow users to create 3D models from Midjourney images with camera movement.

A new storytelling tool is in development, focusing on world-building with a different user interface.

Starting October 1st, purchased fast hours will expire after 60 days.

A new survey focuses on the connection between aesthetics and demographics.

The explor feed algorithms are being tweaked to show a variety of aspect ratios.