3D Optimism | Midjourney Office Hours Recap April 3rd 2024 | Midjourney News
TLDRThe Midjourney Office Hours recap from April 3rd, 2024, highlights a slower progress due to vacations. The team is focusing on the website's new social features, which will initially be tested with a limited number of spaces. Personalization is a work in progress, with challenges due to multiple time zones. An algorithm for improving hand and body text accuracy is being developed, aiming to reduce poor image occurrences. There's a potential for a speed update, but it's contingent on other updates. A caption party is planned to enhance the model's understanding of image-language connections. David hints at a new class of trusted users for rating and captioning, possibly linked to larger rewards. Although video features are not progressing well, optimism remains for a high-quality 3D model in version 7. The feedback leaderboard on the Midjourney website will see more ideas, and the team is considering adding demographic data to understand user preferences better. Consistent character generation might be possible in version 7. The recap ends with a prompt for creating a serene double exposure image, showcasing the art style and inviting followers on Instagram and Twitter.
Takeaways
- 🌐 **Medium Recommendation**: Creative professionals are encouraged to visit Medium for customizable prompts that can save time at work.
- 📅 **Vacation Impact**: Progress has been slower due to people being on vacation.
- 🔧 **Website Development**: The team is working on the website, including new social features, which will be stress-tested with guides and mods.
- 🤝 **Social Spaces**: Initially, there will be a limited number of social spaces focused on quality over quantity.
- 🚀 **Personalization Efforts**: Personalization is a work in progress, with the team operating across multiple time zones, leading to a slower pace.
- 🎨 **Style Random Return**: The 'style random' feature is set to make a comeback, likely through dial tuning, though users won't have access to the tuning.
- 🤖 **Algorithm Improvements**: An algorithm to assist with hands, bodies, and text accuracy is in development, aiming to reduce poor image occurrences.
- 📈 **Performance Update**: A potential speed update could make processes 25-50% faster and cheaper, pending completion of other updates.
- 🎉 **Caption Party**: An upcoming event to teach the version 7 model about the connection between images and language, with possible future rewards.
- 👥 **New User Class**: Mention of a new class of trusted users for rating and captioning, potentially linked to larger rewards.
- 🎥 **Video Updates**: Version 6 model for video is unlikely, but there's confidence in a version 7 model, focusing on high-quality 3D over exportable models.
- 📊 **Feedback Leaderboard**: The website's feedback leaderboard will see more ideas added periodically for community rating.
- 🚫 **Content Policies**: No plans to expand on not-safe-for-workplace features, and no user manipulation of images with the Midjourney model is ready.
- 🧍 **Consistent Characters**: Multiple consistent characters in generation are not available in V6 but may be possible in V7.
- 🎨 **Art Appreciation**: The art in the video was created using a specific prompt and settings, highlighting the capabilities of the Midjourney model.
Q & A
What is the main topic of the mid-Journey office hours recap from April 3rd, 2024?
-The main topic is an update on the progress of the Midjourney platform, including new features, improvements, and upcoming events.
What website is recommended for creatives to check out?
-The website Medium is recommended for creatives, as it sells customizable prompts that can save time at work.
What are the current challenges the Midjourney team is facing in their development process?
-The team is facing slower progress due to people being on vacation and the difficulty of working across multiple time zones.
What new features are being tested for the Midjourney website?
-New social features are being tested, including the creation of public and private spaces, with an initial focus on a low number of spaces with lots of people.
How is the Midjourney team approaching personalization on their platform?
-They are working hard on personalization, but it is moving slower than desired due to the complexity and the distributed nature of the team.
What is the status of the 'style random' feature?
-David mentioned that 'style random' will show up again, likely from dial tuning, but users will not have access to the tuning part.
What improvements are being made to the algorithm for handling hands and bodies in images?
-The team is working on an algorithm to improve the handling of hands and bodies, as well as text accuracy, although it has been finicky and still results in some bad images.
Are there any updates on the speed and cost of the Midjourney platform?
-There might be a small speed update that could make things 25-50% faster and cheaper, but it is dependent on the completion of other updates.
What is the purpose of the upcoming 'caption party'?
-The caption party aims to help teach the version 7 model the connection between images and language. If successful, it might become an official activity with rewards in the future.
What new class of users is briefly mentioned in the recap?
-A new class of users is mentioned who would be trusted with rating and captioning. They might have to qualify for rewards, potentially leading to larger rewards.
What is David's current stance on the 3D model development?
-David is optimistic about having a really good 3D model in version 7, thanks to progress on hardware capture. The focus is on producing high-quality 3D models rather than just exportable ones.
How does the Midjourney team plan to use the feedback leaderboard on their website?
-The team plans to add more ideas to the feedback leaderboard and ask people to rate them. They are also considering adding demographics to understand who is asking for each feature.
Outlines
📝 Mid-Journey Office Hours Recap
This paragraph provides a summary of the Mid-Journey office hours held on April 3rd. It mentions that creative professionals may find Medium, a website offering customizable prompts, useful for their work. The recap highlights that there were no major announcements due to slower progress, likely because team members were on vacation. The main focus is on the website's development, including new social features, which will initially be tested with guides and mods. The team is also working on personalization, although it is progressing more slowly than desired. An upcoming feature is 'style random,' which is expected to return after some tuning. The team is also developing an algorithm to improve the accuracy of hands, bodies, and text in images. They are addressing small pixel artifacts to enhance image quality and are considering a speed update that could make processes 25-50% faster and cheaper. A caption party is planned to help teach the version 7 model about the connection between images and language. There's a mention of a new class of users who might be trusted with rating and captioning, potentially leading to larger rewards. David, presumably a team member, expresses optimism about a good 3D model in version 7 due to hardware capture progress. The feedback leaderboard on the Mid-Journey website is discussed, with plans to add more ideas and possibly incorporate user demographics. The paragraph concludes with a note on the possibility of consistent characters in future versions and a prompt for creating a serene double exposure image.
Mindmap
Keywords
💡creative
💡social features
💡personalization
💡style random
💡algorithm
💡image quality
💡speed update
💡caption party
💡3D model
💡feedback leaderboard
💡double exposure
Highlights
Medium is suggested as a time-saving resource for creative professionals with customizable prompts.
Progress has been slower than usual due to vacations.
The main focus is on the website, including new social features.
Initial social spaces will be limited in number with a focus on quality.
Users will eventually be able to create both public and private spaces.
Personalization is a work in progress, albeit slower than desired.
Style random feature is expected to return, possibly from dial tuning.
An algorithm is being developed to improve hands, bodies, and text accuracy.
Bad images will still occur but with less frequency.
Enhancements are being made to image quality, targeting small pixel artifacts.
A potential speed update could make processes 25-50% faster and cheaper.
A caption party is planned to teach the version 7 model about image-language connections.
There may be a new class of users trusted with rating and captioning.
Video capabilities are not fully satisfactory, with more confidence in a version 7 model.
Focus is on producing high-quality 3D models rather than just exportable ones.
Feedback Leaderboard on the Midjourney website will receive more ideas periodically.
There are no immediate plans to allow user manipulation of images with the Midjourney model.
The possibility of adding demographics to feedback to understand user preferences better.
Multiple consistent characters in generation might be possible in version 7.
A serene double exposure image prompt is shared for artistic inspiration.