Midjourney Updates: What’s New with Image-to-Video, V8, and Multiple References

Better References and Mood Boards Are Coming

Get ready for improved ways to guide your creations. Midjourney plans to roll out an update to SREF (style reference) and introduce something called “mood boards.”

These updates are expected to make references more accurate and useful.

These updates are expected to make references more accurate and useful.

The team is very optimistic about how well they work.

The team is very optimistic about how well they work.

Tests are happening now, with a possible release next week.

Tests are happening now, with a possible release next week.

Working on Multiple Image References

Building on the Omni reference system, Midjourney is training with more data. They are also testing a way to use multiple image references in one go. If this test is successful, they might combine different training efforts into one big update.

Imagine guiding your art with not just one image but several. This could open up many new possibilities for creators.

Speaking of enhancing your workflow, managing multiple references and variations can be time-consuming. Tools like the

Exploring Other Image Quality Boosts

The team is researching other ways to improve image quality and make the art even better. These ideas are promising but are still very early in development. Don’t expect them in the immediate future.

Progress on Video Models

Midjourney is working on video generation. They have successful text-to-video working internally, but they don’t plan to release it as a product right now. Instead, they are focusing on an image-to-video model.

The image-to-video model is currently being fine-tuned.

The image-to-video model is currently being fine-tuned.

The goal is for it to be as good as, or better than, their text-to-video results.

The goal is for it to be as good as, or better than, their text-to-video results.

It’s aiming for dynamic motion and a good balance between quality and how much computing power it uses.

It’s aiming for dynamic motion and a good balance between quality and how much computing power it uses.

They hope to finish training everything by the end of this month and then begin testing.

They hope to finish training everything by the end of this month and then begin testing.

This first video model might have some initial limitations, as they already see ways to improve it in future versions. However, they are optimistic about this initial release.

Version 8 and V7 Speed

A new major image model version, V8, is in the works. It’s expected to be a significant upgrade, combining new data with a new model design. They will likely move to V8 instead of adding every new feature to V7.

Meanwhile, they are trying to make V7 faster before it becomes the standard default model. They need to optimize it to ensure they have enough computing power for all users. They’re even testing modes faster than “draft” mode, although feedback on using voice prompts for them wasn’t ideal right now.

Automating repetitive tasks is key to getting more out of powerful tools like Midjourney V7 and the upcoming V8. Consider using the

New Interfaces and Secret Projects

Midjourney is exploring a new professional interface concept that might require a high-quality display for the best experience. They are also thinking more about how Midjourney works on mobile devices.

And yes, they are making progress on other “secret projects” too, though details remain under wraps.

Try a Shared Prompt

In the original office hours recap, a prompt for the art shown was shared. It’s a mood board combination using stylized 500. Here are the codes:

Type “-p “

Type “-p “

Followed by “VQR BR V”

Followed by “VQR BR V”

Then “LF1 D U W6”

Then “LF1 D U W6”

And finally “YQ Z 3T TDO”

And finally “YQ Z 3T TDO”

These examples were created by community members.

Keeping track of prompts and variations can be tricky. The

Looking Ahead

It sounds like Midjourney is busy with many exciting developments. From better reference tools and the possibility of multiple image references to video capabilities and a future V8 model, there’s a lot to look forward to.

Stay tuned for these updates as they are rolled out!