🚀 Midjourney Office Hours Highlights: April 9th

Midjourney has launched the alpha version of its V7 model, kicking off a series of updates aimed at making V7 the default. New features like tile support and a remix function are on the way, along with improvements in speed and image quality. User feedback is crucial, with new tools for rating images and "ranking parties" to gather input. Midjourney is also working on video and 3D model advancements, alongside major infrastructure upgrades to enhance performance. Stay tuned for more updates and some exciting secret projects!

🚀 V7 Alpha Model Release and Updates

Midjourney has unveiled the initial alpha version of its V7 model, which users can access by enabling it explicitly. This marks the beginning of a series of weekly updates planned over the next one to two months, with the goal of making V7 the default model. The immediate next release, expected today or tomorrow, will introduce tile support, a remix feature, and an enhanced mode for drafts, promising a more versatile and user-friendly experience.

🗣️ V7 User Feedback Integration

To refine the V7 model, Midjourney has introduced new feedback buttons on the website's Lightbox, allowing users to rate images and contribute to the model's improvement. This initiative underscores the importance of community feedback in shaping future updates. Additionally, Midjourney is organizing "ranking parties" to further engage users, with the next event likely scheduled for this Friday and possibly the following week.

⚠️ Caution Using User Data for Model Training

Midjourney acknowledges the risks associated with using user data for model training, particularly in avoiding unintended extremes like overly smooth skin or exaggerated physiques. The team is committed to carefully balancing data usage to prevent such outcomes while aiming to increase the frequency of aesthetically exceptional images from the current 1/10 to approximately 4/10.

⏩ Model Speed and Quality Optimizations

Significant speed improvements are in the pipeline, with the potential to double the current speed and add an additional 20%, possibly achieving another 2x increase. While there may be minor trade-offs in image quality, the goal is to maintain zero loss. Midjourney is also exploring higher-quality modes (Q4/Q8) to enhance image coherence and detail, ensuring users receive the best possible output.

🔧 Upcoming Feature Priorities

The "Omni Reference" feature is expected to roll out within two weeks, initially in a limited release to gather feedback and prioritize enhancements. Improved upscalers and image quality enhancements are actively being developed, with a strong commitment to iterative updates based on real user feedback for rapid improvements.

🎥 Video and 3D Model Development

Training for video models is underway, with completion anticipated by the end of the month. However, the release timing will depend on further refinement. Once the V7 and video milestones are achieved, Midjourney plans to resume its dedicated focus on 3D model development, promising exciting advancements in this area.

💻 Infrastructure Upgrades

Major server infrastructure upgrades are ongoing and will continue for the next eight months. These upgrades aim to enable dramatically faster and higher-quality models, with noticeable improvements expected by the end of the year. Users can look forward to real-time fast models and higher-quality standard models, significantly surpassing current capabilities.

🤫 Secret Projects Teasers

Midjourney is testing two hardware projects, one wearable and one immersive, with announcements expected soon. While these longer-term projects are in development, the most immediately available product announcement has been delayed until it is ready for order, keeping users eagerly anticipating what's next.

TL;DR

🚀 V7 Alpha Model Release and Updates: Midjourney launches V7 alpha with weekly updates planned to make it the default model soon.

🗣️ V7 User Feedback Integration: New feedback buttons and "ranking parties" to enhance V7 through community input.

⚠️ Caution Using User Data for Model Training: Balancing data usage to avoid extremes, aiming for more exceptional images.

⏩ Model Speed and Quality Optimizations: Speed improvements underway, exploring higher-quality modes with minimal quality trade-offs.

🔧 Upcoming Feature Priorities: "Omni Reference" feature coming soon, with ongoing upscaler and quality enhancements.

🎥 Video and 3D Model Development: Video models in training, 3D model focus to resume post-V7 and video milestones.

💻 Infrastructure Upgrades: Major server upgrades for faster, higher-quality models, with improvements expected by year-end.

🤫 Secret Projects Teasers: Testing wearable and immersive hardware projects, with announcements on the horizon.

Stay tuned for more updates as Midjourney continues to innovate and enhance its platform!

If you want to support me, feel free to buy me a coffee ☕️ 

If you’re not subscribed yet to my newsletter Imagine Weekly, I’d be thrilled to welcome you on board!