Midjourney has never stopped updating its features. The latest Character Reference, Style Codes, and web-based Image Editor make the creative process more intuitive than ever. If you're still on an older version or used to typing parameters manually in Discord, these new additions are worth exploring.
Character Reference and Style Reference: No More Manual Tweaking
Character Reference now supports inputting multiple characters at once. The official update has enabled the Multiple CREFs option. You can upload several photos of the same person from different angles, and Midjourney will automatically learn the facial features. The generated images are much more consistent in character identity compared to earlier versions. Style Reference has also been upgraded. In addition to the original --sref code, it now integrates the Style Codes feature. You can simply copy a short code to lock in a specific art style, saving you the hassle of adjusting style values manually every time.
Web Alpha and Personalized Preferences
The web version is no longer just an accessory to Discord. The editing capabilities have expanded from simple Vary (variations) to a full Image Editor, including Reframe for recomposing and Repaint for partial redrawing. The personalization preference system (the --p parameter) has also evolved. You can now train the AI by rating images you like or dislike, making it better align with your aesthetic. This system has become notably more effective since V6.1—the AI is less prone to misinterpret complex prompts.
Image Editor and V8.1 Accelerated Rendering
The current default version is V7, but V8.1 is the fastest new version, with rendering speeds approximately 4–5 times faster than the older version and native output of HD 2K images. The Raw Mode option is also available again, ideal for users who want to preserve more original details. The Patchwork "world-building" tool is set to launch by the end of the year, enabling multi-image interconnected scenes—a huge boost for game or concept design workflows.
