While V7 improves coherence in image details like textures and objects, some features such as image upscaling and retexturing are not yet available but are expected within two months. Midjourney CEO David Holz describes V7 as a "totally different architecture" and encourages users to experiment with different prompting styles to explore the model's strengths and weaknesses.
Key takeaways:
- Midjourney's V7 image model features a new architecture with improved text prompt handling, higher image quality, and default personalization based on user-rated images.
- Users must rate around 200 images to create a personalization profile, which tunes the model to their visual preferences, and can toggle V7 on or off via the Midjourney website or Discord chatbot.
- V7 is available in Turbo and Relax modes and introduces Draft Mode, which renders images faster and cheaper but at a lower quality, with options for enhancement.
- Some features like image upscaling and retexturing are not yet available in V7 but are expected to be added within two months.