🚀 Midjourney Office Hours Highlights: December 4th

As Midjourney gears up for an exciting year ahead, the much-anticipated V7 model is slated for a January 2025 release, promising enhanced quality through optimized text dataset balance and new aesthetic and personalization systems. While the team navigates the challenge of maintaining visual output quality, they are also exploring the future of video models, weighing community preferences for affordability versus premium quality. Upcoming features like Personalization Profiles and Storytelling Tools are set to enrich user experience, despite potential capacity challenges at launch. Meanwhile, Midjourney's shift towards a modular architecture aims to decouple personalization and datasets from model versions, allowing for greater flexibility and scalability in response to user demands for higher quality options.

For a quick overview, head straight to the TL;DR; section otherwise keep on reading.

Extended Recap

📊 V7 Model Status

The much-anticipated V7 model is set for a January 2025 release, as the December launch is no longer feasible. The team is making significant progress in model training, focusing on optimizing the balance of text datasets. A key challenge is ensuring that while more text can enhance quality, it doesn't compromise visual output. Development is also underway for V7.1 and future versions, which will introduce new aesthetic and personalization systems. The plan includes slower but higher-quality model variants, similar to the Q2/Q4 approach, in response to user demands for superior quality options.

🎥 Video Model Strategy

A crucial decision point has emerged regarding the video model: should it be affordable or premium? A community poll revealed a 50/50 split in preferences. Three potential strategies are being considered:

• An affordable model with lower performance.

• A premium model offering the highest quality.

• A limited-scope model balancing cost and performance.

Business constraints are significant, as current video models require substantial investment, with ROI challenges due to Midjourney's self-funded nature. Technology limitations also make it difficult to deliver affordable, high-quality video. Plans are in place to collect user rankings for video model training.

📅 Upcoming Releases

Exciting new features are on the horizon:

• Personalization Profiles: Support for multiple personalization options and a new "mood board" feature that allows personalization from a collection of images.

• Storytelling Tools: The first experimental version will be demoed at the next office hours, though potential capacity issues are anticipated at launch.

🔮 Future Development

Enhancements in personalization are a priority, with improved character references for V7 and a new "omni reference" system in development. This will enable specific object and design integration. A big batch release is planned, with decisions pending on whether it will coincide with the December release or be integrated with V7. The final approach for the video model will also be selected, with a release schedule and prioritization set for 2025.

🏗️ Architectural Changes

Midjourney is moving towards a modular architecture, which involves:

• Separating personalization from model versions.

• Decoupling datasets from model versions.

• Allowing for independent scaling and architectural modifications.

These changes aim to enhance flexibility and scalability, ensuring that Midjourney can continue to innovate and meet user needs effectively.

TL;DR:

📊 V7 Model: January 2025 release; focus on text dataset balance and quality.

🎥 Video Model: Decision between affordable vs. premium; significant investment needed.

✨ New Features: Personalization profiles, mood board, and storytelling tools coming soon.

🔮 Future Plans: Enhanced personalization, omni reference system, and modular architecture.

🏗️ Architectural Changes: Midjourney is transitioning to a modular architecture by separating personalization and datasets from model versions, enabling independent scaling and modifications.

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!