🚀 Midjourney Office Hours Highlights: July 16th

Midjourney’s latest office hours revealed major progress on the V8 model and new creative tools, signaling a strong focus on quality and innovation. Even with summer slowdowns, the team is pushing forward on features that promise to enhance both user experience and creative possibilities.

V8 Model Development 🚀

V8 has become the central focus for Midjourney’s technical efforts, with early internal testing already underway as of July. The team plans extensive training runs over the next 2–3 months, aiming for significant leaps in text rendering, scene composition, and overall image quality. Notably, V8 will be partially backward-compatible with V7, enabling some features to be released ahead of the full launch. If initial results are promising, the team intends to refine V8 for several months rather than moving quickly to a V9 iteration, prioritizing stability and quality.

Video Features and Roadmap 🎬

Significant progress has been made on video features, including looping and start/end frame controls, which are now complete and in testing. A high-resolution video mode (likely 720p) is built but remains expensive to run, so initial trials will focus on professional users. The video upscaler is still in early development and months away from release. Turbo mode, offering faster video rendering, is now live but is also resource-intensive and targeted at professionals. Additionally, a post-processing sound effects layer for videos is being considered. Some delays in video feature rollout are attributed to ongoing compute cluster deployment issues.

V7.1 Possibility

There is a possibility of a V7.1 release emerging as a byproduct of the V8 training process, depending on the compatibility and performance of early V8 components. However, this is not guaranteed and is described as a hopeful intermediary step rather than a planned milestone.

Omnireference & Character Consistency

Efforts are underway to enhance datasets for improved character consistency across different poses, outfits, and settings. These improvements are not expected to appear in V7, but are more likely to be incorporated into V8 or later versions, reflecting a longer-term commitment to addressing this challenge.

Mega Model & Scaling Goals

Looking ahead, the development of even larger “mega models” (V9, V10, V11, etc.) will depend on securing new partnerships or significantly increasing compute resources. Current models are trained with substantially less compute than industry leaders, and scaling up is seen as a key lever for boosting performance and competitiveness.

Style Explorer & Mood Boards 🎨

The Style Explorer tool is in active development, with two competing concepts currently under review. Mood boards have received silent updates to improve their performance and weighting behavior. There is also consideration for a dedicated mood/style tab to enhance discoverability and user engagement with these creative tools.

Enterprise API Investigation 🏢

Midjourney is officially exploring the development of a commercial API, having launched a public survey to assess demand and gather contacts. A second, more detailed survey is planned to better understand the specific needs of companies of all sizes. Notably, the Twitter announcement uses a separate form to mitigate spam.

Discovery & Recommendations 🔍

Work is ongoing to build content recommendation features aimed at improving image and video discovery. The goal is to create a more personalized and engaging creative experience for users, making it easier to find relevant and inspiring content.

Secret Projects & Internal Demos 🕵️‍♂️

Several stealth and experimental projects are currently in development. The upcoming team offsite will feature internal demos, serving as preparation for future public announcements and releases.

Storytelling Labs 📚

More story-driven Labs experiments are on the way, intended as creative testbeds rather than official products. The team has noted some user confusion regarding the purpose and status of these Labs, and may clarify their experimental nature moving forward.

Team, Timeline & Operations

July has been a slower month due to team vacations and post-launch recovery. Some staff are traveling to Europe for meetings, and work continues on future editor tools, which are likely to integrate with V8. Operational pace is expected to pick up after this period of rest and regrouping.

TL;DR

• 🚀 V8 Model Development: V8 is now the main focus, with early testing and major improvements planned; backward compatibility with V7 will allow some features to roll out early.

🎬 Video Features and Roadmap: Looping and frame controls are in testing; high-res video and turbo mode are available for professionals, with further enhancements and upscaling in progress.

🌀 V7.1 Possibility: A V7.1 release may occur if early V8 components are compatible, but this is not guaranteed.

🧑‍🤝‍🧑 Omnireference & Character Consistency: Improved character consistency is in development, targeted for V8 or later.

📈 Mega Model & Scaling Goals: Future mega models depend on increased compute power and partnerships, aiming for industry-leading performance.

🎨 Style Explorer & Mood Boards: New style exploration tools and improved mood boards are being developed to enhance creative workflows.

🏢 Enterprise API Investigation: Midjourney is surveying demand for a commercial API, with further research planned to address business needs.

🔍 Discovery & Recommendations: Personalized content recommendations are being built to improve discovery and engagement.

🕵️‍♂️ Secret Projects & Internal Demos: Experimental projects are underway, with internal demos planned for the next team offsite.

📚 Storytelling Labs: More Labs experiments are coming, serving as creative testbeds rather than official products.

🗓️ Team, Timeline & Operations: July is slower due to vacations and recovery, with ongoing work on editor tools and European team meetings.

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!