Guide to Midjourney’s Latest Updates: Character Refiners, Styles, and More

Midjourney keeps adding new features. Recently, they rolled out exciting updates. These include better ways to handle styles, an improved tool to get text from images, and a new character reference feature. Let's look at what these updates mean for creating images.

Understanding Midjourney’s Style Reference (--sref)

Style reference lets you copy the artistic look from one image and apply it to new images you generate. This includes colors, art medium, and the general feel.

Midjourney now has four versions of style reference:

• Version 1: The first release.

• Version 2: Became the default after version 1.

• Version 3: An improved version based on Version 1.

• Version 4: An improved version based on Version 2 and the current default.

You can pick a specific version using the parameter --sv followed by the version number in your prompt. If you don't add --sv, Midjourney uses the default version (currently V4).

These new versions were made to fix two common issues. First, they improve performance. Older versions sometimes brought elements like people or objects from the reference image into the new image, not just the style. The new versions should do this less.

Second, they improve how the style weight parameter (--sw) works. In older versions, the style effect hit its maximum impact quickly, even at lower weight values. The newer versions show a more even change in effect as you increase the weight, at least at lower weights.

Many users find version 4, the default, gives results that stick closer to the reference image's style. Version 3 can create results that are more creative or interesting, even if they stray a bit from the original style.

Managing Midjourney features and parameters can get complex quickly. Tools designed for automation can help streamline your workflow. Consider using a tool like the Midjourney Automation Suite from TitanXT to simplify working with parameters like --sref.

Using the Image-to-Text feature: Describe

Midjourney has a feature you might not know about: `/describe`. It takes an image and gives you four text descriptions that you can use as prompts. This is great when you need ideas or aren't sure how to describe an image you like.

The `/describe` feature has been updated to work better with Midjourney V6 results.

To use it, type `/describe` in your chat, add a space, and then either paste an image link or upload an image file. Press Enter. Midjourney will give you four possible text prompts based on that image.

These descriptions are like starting points. They won't make an exact copy of your original image. But they show you words and phrases Midjourney sees as important in that image. Words that appear in multiple descriptions are often powerful descriptors.

You can use `/describe` to find inspiration from images you admire or to see how Midjourney interprets visual elements. If you're looking for style words, you might also try that image as a style reference (--sref) first.

Remember you can also use the `/shorten` command with the text results from `/describe` or any prompt to see which words Midjourney considers most important.

Introducing the Character Reference (--cref) Feature

A big new update is the character reference feature. This works like style reference, but it tries to match a character from a reference image instead of just the style. This is very useful for authors or creators who need the same character to appear in many different images or scenes.

To use it, write your prompt, add `--cref`, and then include a URL for the image of the character you want to reference. Midjourney will try to create that character in your new image.

Character reference works best with images made by Midjourney and characters with clear, unique features. It won't copy small details exactly, like freckles or shirt designs, but it aims to keep the main look of the character.

A cool way to use --cref is to put a character into different art styles or photo looks. You can reference a character image and then prompt for different styles like "pop art" or "Kodak Portrait 400 film" without needing to describe the character in the text prompt.

You can control how strong the effect is with the character weight parameter `--cw`. This goes from 0 to 100, and 100 is the default. When --cw is 100, Midjourney pays attention to the character's face and clothing. When --cw is 0, Midjourney focuses only on the face, letting you change things like their clothes while keeping the same face.

It's important to know that Midjourney will *always* try to keep the face from the reference image, no matter what the character weight is. So, `--cw` is mostly for controlling if the clothing is copied.

Character reference is still new and learning. Sometimes you might get unexpected results, like a character facing the wrong way. You might need to adjust your text prompt to help Midjourney. For example, changing "walking through a field" to "standing in a field" and adding "looking at the camera" can help ensure the character faces forward.

Staying on top of new Midjourney features like character reference can be time-consuming. Streamline your process with a tool like the Midjourney Automation Suite from TitanXT. It can help you work more efficiently with these advanced parameters.

Summary

Midjourney keeps adding powerful tools. The updates to style reference give more control, the image-to-text feature is better for V6 prompts, and the new character reference helps create consistent characters across different images. These features give creators more options for bringing their ideas to life.

To make the most of these features and manage your Midjourney workflow effectively, explore automated solutions. The Midjourney Automation Suite from TitanXT is built to help you streamline your image generation process.