Explore New Midjourney Features: OREF, V7 Boosts, and Editor Changes

Check out the latest updates from Midjourney in May. There are some big changes that can help you create better images.

Get Started with Omni Reference (OREF)

The biggest new feature is Omni Reference, or OREF. You can use OREF with the V7 model. Unlike Character Reference in V6, which just worked for people, OREF is much more flexible. Use it for characters, objects, scenes, logos, and more.

Make sure you have V7 selected in your Midjourney settings. Then, go to the prompt bar on the website. Drag an image into the bar. You will see a spot that says "Omni reference" where you can drop it.

There is an OREF weight parameter, `-w`. This helps you control how similar your new images are to the reference image. A lower value means less similarity, a higher value means more.

You can even use OREF for more than one character now. The Midjourney team is working on better ways to do multiple consistent references. It is not fully ready yet. It might come in a future V7 update or possibly V8.

Working with features like OREF can involve trying many variations. To make this easier, consider using the TitanXT Midjourney Automator. It helps manage your jobs so you can focus on creating.

Midjourney V7 Gets Faster and Better Quality Options

V7 has received several updates. Fast mode is now available for V7. Midjourney also made speed improvements throughout the month. V7 is now about 40% faster than before.

There are new quality settings you can try: Q2 and Q4. The basic quality setting is 1. Using a higher number tells Midjourney to spend more time and effort on your images. This can give you results with more detail and higher quality.

Keep in mind that Q2 and Q4 jobs cost more GPU hours. If you use Fast or Turbo mode, watch how many hours you use.

It looks like we probably will not get a V7 upscaler. We may have to wait for V8 for that.

A new SREF (Style Reference) and Mood Board system is coming soon. It is expected to look very good and be more accurate. Midjourney is also building a Style Explorer tool. This will help you find and use SREF codes more easily.

V7 will become the standard model very soon. Midjourney plans at least one more aesthetics update for V7. There is a small chance we could see a 7.1 update depending on how some tests go. After V7 becomes the default, they will spend a few months getting ready for V8.

Updates to the Midjourney Editor

Midjourney made a small change to the editor after getting feedback. When you click "Edit" on an image, you now see a simpler editor view. It looks like the older basic editor.

In this basic editor, you can change the aspect ratio, erase parts of the image, and change your prompt.

You can also rotate your image now. Hover near a corner and click and drag. There is a small issue though. If you rotate in this basic editor, the rotation might not show up in the final results you get.

Rotation does work correctly in the full editor. To get to the full editor, click "Open in edit tab." The full editor also has features like layers and retexture.

You can go back to your starting image by clicking the "X" button. There are "Vary" and "Upscale" buttons in the basic editor view, but it is not clear why. If you edit the image here and then click Vary or Upscale, those edits are not used. Vary and Upscale only work on images already created.

Midjourney Gets Ready for Video

Midjourney will start testing its video model with guides and mods soon. Both text-to-video and image-to-video are working well. They still need to optimize things.

They have said the video quality, motion, and overall look are promising.

The first video model release will be simple. Midjourney plans to add more features later. They want to offer different quality levels, speeds, and resolutions.

The goal is for all subscription plans to be able to use video. The cost in GPU hours will change based on the settings you pick.

There will be some small video rating parties. This will help with the final steps of development.

New servers are coming online over the next few weeks. These will be needed before video can launch. Video will likely launch without Relax mode at first. They need to understand how much computing power everyone needs.

Midjourney hopes to put out its first video model this month.

Getting consistent, high-quality results, whether images or soon video, requires understanding many settings. Using a tool like the TitanXT Midjourney Automator can help you manage testing and generating at scale.

You can help Midjourney decide which features to build next. Check out their updated voting system at midjourney.com/ideas. You can see what features people are asking for the most.

Simplify your workflow whether exploring OREF, V7 settings, or preparing for video. The TitanXT Midjourney Automator is built to make managing your Midjourney tasks easier.

Summary

May brought useful Midjourney updates. We got the new OREF reference feature and improvements to V7 that make it faster with new quality options. The editor received a small update based on feedback. Testing for the first video model is starting, with a release expected soon.

Stay tuned for more updates as Midjourney continues to add features.