V8 Alpha
By Steven Van ·
Hi everyone! Today we’re letting the community test an early version of our V8 model on alpha.midjourney.com. We want your feedback and your help.
Midjourney has opened an early version of its V8 image model for alpha testing at alpha.midjourney.com, and is asking the community to try it out and send feedback in the announcement.
The company says V8 follows detailed prompts more closely, understands personalization, style references and moodboards better, produces more coherent and detailed images, and renders text placed in quotes more reliably. Image generation is about 5x faster than before, and the web interface has been upgraded to keep up with the speed.
- A new Grid Mode for focusing on a single large set of images, an improved conversation mode, and settings moved into sidebars so they no longer block the view.
- Support for multiple aspect ratios and the --chaos, --weird, --exp and --raw parameters, plus backwards compatibility with V7 personalization profiles, moodboards and style references.
- A new --hd mode that natively renders images at 2K resolution, and a --q 4 mode for extra coherence.
Relax mode isn't supported yet. For now, --hd, --q 4, sref and Moodboard jobs run four times slower and cost four times as much as regular jobs. Midjourney is asking testers to rate V8 images in the lightbox to help guide improvements, and suggests switching to --raw mode or using moodboards and srefs for a plainer, photo-style look, since V8 leans toward heavy stylization and longer, more specific prompts by default.
