3D Optimism | Midjourney Office Hours Recap April 3rd 2024 | Midjourney News
TLDRIn the mid-April Journey office hours recap, updates on the platform's development were shared. Despite a slower progress due to vacations, work continues on new social features, personalization, and improving text and image accuracy. An upcoming caption party aims to enhance the connection between images and language. Future plans include a new class of trusted users for content moderation and potential speed improvements. Hardware advancements promise high-quality 3D models, and there's a tease about potential new features based on community feedback.
Takeaways
- π Medium is recommended for creatives as a customizable prompts website that can save time.
- ποΈ Progress has been slower due to vacations, but the team is working on new social features for the website.
- π€ Testing of social features will start with a limited number of spaces to stress test the system before public access.
- π¨ Personalization is being improved, albeit at a slower pace due to multiple time zones and complex development.
- ποΈ Style, random feature might return, but without access to the tuning part.
- π€ An algorithm is being developed to improve the accuracy of hands, bodies, and text in images.
- πΌοΈ Efforts are being made to enhance image quality and reduce pixel artifacts.
- π A small speed update is planned, making processes 25-50% faster and cheaper.
- π A caption party is upcoming to help teach the version 7 model about the connection between images and language.
- π A new class of users may be introduced for rating and captioning, potentially with rewards in the future.
- π₯ Video features are still in development, but version 6 model may not include it.
- π High-quality 3D models are being focused on, with less emphasis on exportable 3D for the time being.
Q & A
What does the speaker recommend for employed creatives?
-The speaker recommends that employed creatives check out Medium, a website selling customizable prompts, as it is easy to use and might save them time at work.
What is the current status regarding the office's progress?
-The progress has been slower than usual due to people being on vacation. The main focus is on the website, including new social features, which will be tested with guides and mods.
What can users expect from the new social features?
-Initially, there will be a low number of social spaces with lots of people, allowing them to stress test the system. Eventually, every user will be able to create public and private spaces.
How is the team addressing personalization?
-The team is working hard on personalization, but it is moving slower than desired due to having people working across multiple time zones, which makes it difficult to progress quickly.
What is the status of the 'Style Random' feature?
-The 'Style Random' feature will show up again, but it is not clear what it will be. It seems like it will come from dial tuning, but users won't have access to the tuning part.
What improvements are being made to the hands and bodies algorithm?
-The team is working on an algorithm to improve the accuracy of hands and bodies as well as text. They believe it will work but acknowledge it has been finicky.
Are there any updates planned for image quality?
-Yes, they are working on improving image quality, particularly for small pixel artifacts. They believe they have a way to significantly enhance it.
What about the potential speed update for the system?
-There might be a small speed update making things 25-50% faster and cheaper. However, this update will be released after completing other updates.
What is the 'Caption Party' and its purpose?
-The 'Caption Party' is an upcoming event aimed at teaching the version 7 model the connection between images and language. Initially, it will be a test, but if successful, it might become an official activity where users can earn rewards in the future.
What new user class is being considered?
-A new class of users is being considered, consisting of trusted individuals who would be responsible for rating and captioning. Users might need to qualify for these rewards.
What is the speaker's stance on 3D models?
-The speaker is optimistic about having a really good 3D model in version 7, thanks to the progress made on hardware capture. However, the focus will be on producing high-quality 3D rather than exportable 3D models.
What does the speaker say about consistent characters in a generation?
-The speaker mentions that multiple consistent characters in a generation will not be available in version 6, but it might be possible in version 7.
Outlines
π’ Mid-Journey Office Hours Recap for April 3rd
The recap starts with a recommendation for creatives to check out Medium, a website for customizable prompts. It mentions that progress has been slower due to vacations, and the main focus is on the website's new social features, which will be tested with guides and mods. Initially, there will be a limited number of social spaces. Personalization is also being worked on, albeit at a slower pace due to the involvement of people across multiple time zones. Style and random features are mentioned, with an algorithm being developed to improve text accuracy for hands, bodies, and overall image quality. A speed update is also anticipated, but it's contingent on completing other updates first. An upcoming caption party aims to improve the connection between images and language for the version 7 model. A new class of trusted users for rating and captioning is being considered. Video improvements and 3D model developments are also discussed, with a focus on quality over exportability. Lastly, the feedback leaderboard on the Mid-Journey website is mentioned, with plans to add more ideas and potentially incorporate user-requested features.
Mindmap
Keywords
π‘Medium
π‘Social Features
π‘Personalization
π‘Algorithm
π‘Image Quality
π‘Speed Update
π‘Caption Party
π‘User Rewards
π‘3D Model
π‘Feedback Leaderboard
π‘Consistent Characters
Highlights
Medium, a website selling customizable prompts, is recommended for employed creatives to save time at work.
Progress has been slower than usual due to people being on vacation.
The main focus is on the website, including new social features to be tested with guides and mods.
Initially, there will be a limited number of social spaces with a focus on stress testing the system.
Personalization is being worked on, albeit at a slower pace due to multiple time zones.
Style, random will make a return, but without access to the tuning part.
An algorithm is being developed to improve hands, bodies, and text accuracy.
Bad images will still occur, but less frequently with ongoing improvements.
Efforts are being made to enhance image quality and reduce small pixel artifacts.
A potential small speed update is in the works, making processes 25-50% faster and cheaper.
The speed update release is contingent on completing other updates first.
A caption party is planned to help the version 7 model learn the connection between images and language.
The caption party may become an official activity with rewards in the future.
A new class of users may be introduced, trusted with rating and captioning responsibilities.
Video features are being considered, but a version 6 model is unlikely.
Version 7 model is anticipated to have impressive 3D capabilities thanks to hardware capture progress.
High-quality 3D production is prioritized over exportable 3D features.
The feedback leaderboard on the Mid Journey website will receive more ideas periodically for community rating.
The team is not planning to add not-safe-for-workplace features.
Demographics might be added to the feedback system to understand feature requests better.
Multiple consistent characters in a generation may be possible in version 7.
A serene double exposure image prompt is shared, showcasing the use of stylize and chaos settings.
The speaker's social media handles are provided for following their work on Instagram and Twitter.