Midjourney New Features Deep Dive: Personalized Settings & Image Editor Full Upgrade

Midjourney's recent updates have taken AI image generation to the next level. With features like personalization codes, a brand-new Image Editor, and character or style references, creators can now fine-tune outputs with greater precision. This guide breaks down the key improvements so you can make the most of these tools — no more guessing or relying on templates.

Personalization Settings & Personalization Codes

Midjourney introduces Personalization, allowing you to save your aesthetic preferences with a personalization code. Simply add the --p parameter to your prompt, and the system adjusts outputs based on your past likes or specified style. This means each generated image aligns more closely with your taste, without constant parameter tweaking.

Additionally, the official Moodboard feature lets you upload reference image sets. The model learns from them and generates a consistent style across a series of images — especially useful for designers who need brand visual uniformity.

Image Editor: Reframe & Repaint

The new web-based Image Editor transforms post-generation adjustments. Without leaving the page, you can Reframe, Repaint, or make localized edits to your generated images. For example, to reposition a subject or fix hand details, simply select the area and enter a new instruction — the model intelligently fills in the changes.

The editor also integrates Vary and Upscale features. After generating an image, you can quickly create multiple variations or use the upgraded upscaler to enhance texture quality. Standard task generation is now about 25% faster, saving you significant waiting time.

Character Reference & Style Reference

Maintaining character consistency has always been a challenge in AI art. The new Character Reference (--cref parameter) lets you upload a person's image, and all subsequent generations preserve that character's facial features. Even better, you can reference multiple characters (Multiple CREFs) to easily create multi-person scenes.

Style Reference (--sref parameter) allows you to specify an artistic style or texture. The official site also shares a library of SREF Codes — you can directly apply pre-tuned aesthetic templates like watercolor, cyberpunk, or 8-bit pixel art. Just add --sref <code> at the end of your prompt, and the style transfers instantly.

Image Quality & Detail Improvements

The latest model shows significant improvements in image coherence. Arms, legs, hands, and other body parts are no longer distorted. Background characters' faces and small details are crisper. Texture, skin quality, and 8-bit retro effects have all been enhanced. For extreme detail, enable --q 2 mode — it increases generation time by about 25% but delivers much richer image quality.

Text rendering accuracy has also improved noticeably. Simply enclose the desired words in quotes within your prompt, e.g., "NEW", and the generated image will display the text more precisely — a huge help for poster or logo design.