Last Christmas, Midjourney fans got a nice gift with the surprise release of Midjourney V6. I still remember waking up to the news and spending all morning just trying out prompts in Discord.
But since then, Adobe Firefly came out with a new model, Meta released their own AI image generator, and OpenAI started teasing Sora. Midjourney, on the other hand, focused on releasing new features and getting their web application up and running. The community has been satisfied with all these. But after months of silence about Midjourney 7, some of them are getting antsy.
So, let’s talk about it. Here’s everything we know about Midjourney 7 so far, including possible new features and when it’s going to be released.
What We Know So Far
Truthfully, we don’t know much. Midjourney has been eerily quiet about Midjourney 7 the past few months, which is highly unusual. But it could also be taken as a sign that something big is coming. Based on previous updates and Midjourney’s latest community posts, there are some things we can be confident that are coming. These are:
Improved Creativity
There are two things that Midjourney always improves on every model update. The first is creativity. For context, here is a quick gallery of Midjourney-generated images to show how they evolved from V1 to V6 using the same prompt.
Prompt: a girl in a garden, earth colors, dreamlike realism, honeycore, Casey Weldon, Satoshi Kon, minimalism, topographical realism, magic realism, whimsical and fluid illustrations with multilayered realism featuring surrealistic elements inspired by graphic design, abstraction
The evolution of Midjourney V1 to V3 focused more on attaining structural integrity, while the last three honed down on creativity. We can expect Midjourney V7 to do the same as the latter, creating better and more intricate images than V6.
Improved Nuance
Another struggle of AI image generators is nuance, and Midjourney is no exception. To their credit, they’re likely one of the few models (DALL-E 3 included) that can understand complex prompts, but they’re also still far from perfect.
Midjourney 7 could be a total game-changer in terms of nuance. If not, at the rate they’re improving, they’re still poised to take over DALL-E 3. This is due to the fact that OpenAI has started focusing on multimodal prompts with GPT-4o and text-to-video with their newest model: Sora.
Better Consistency and Style Control
Over the last couple of months, most of Midjourney’s improvements focused on one aspect: consistency. These changes include better control not just for art style, but also characters. This is implemented through these parameters:
- Style Reference: Allows you to upload an image and Midjourney imitates its style and vibe using the –sref parameter. You can also adjust the style strength using the –sw keyword.
- Character Reference: Works similar to Style Reference, but instead of copying the reference’s style, it works for the image’s subject.
Midjourney has also made steady improvements to both of these features in terms of consistency and creativity. In my opinion, these features (and by extension, listening to the community) are the reasons why Midjourney is still the best AI image generator by a mile.
As good as Midjourney is now, there are still some points of improvement that they need to address. Fortunately, the company behind it is actively listening to the community. They even started a page where users can submit improvement ideas and vote on what they want to see the most.
Here are some of the most upvoted ones and the ideas I found most interesting:
- Multiple Angles. This would allow users to generate the same image but from different angles. A feature like this would work wonders in creating mockups and movie scenes.
- Image Editing. A web-based editor for uploaded images. I imagine this would work like the inpainting and outpainting feature of Adobe Firefly with Photoshop.
- Transparent Background. Bypasses the hassle of using third-party applications to remove background.
- Color Palette Control. This would give users better control on the generated images’ color palette.
- Font Control. Along with DALL-E 3 and Adobe Firefly 3, Midjourney is one of the few AI image generators that can generate text to a considerably consistent degree. Having more control over the font it uses would save users the hassle of editing with Photoshop.
- Sketch-to-Image. This will give users the option to upload a sketch and Midjourney would use it as a blueprint to create the final image.
- 3D Rendering. Create 3D objects using Midjourney only.
- Held Objects, Hands, and Fingers. Midjourney does structural integrity just fine, but it has always struggled with fingers and held objects (especially swords).
When Will Midjourney 7 Be Released?
Again, we don’t actually know when Midjourney 7 will be released. The team has been extra tight-lipped about this new model so there are many who’s wondering if they’re even planning to release it this year or if they’re just focusing on improving Midjourney V6 and Niji.
But to answer this question, we have to look at past information. Midjourney V1 was released in February 2022, V2 in April 2022, V3 in July 2022, V4 in November 2022, V5 in March 2023, and V6 in December 2023. As the models evolve, the wait times for the next one becomes bigger.
So, given that the gap between V5 and V6 is 9 months, it’s reasonable to think that Midjourney V7 won’t come out until September 2024 at the earliest. Popular social prediction platforms such as Manifold Market estimates that the likeliest time Midjourney V7 will be released is around October or early November.
The Bottom Line
Estimating when Midjourney V7 will come out is a guessing game at this point. The company behind it isn’t really giving us any teaser about it, which makes sense considering that the release of V6 last Christmas was also a surprise.
But if I were to guess? I’d say it’s coming in October, but don’t take my word for it.
What we can expect, however, is that this might be one of Midjourney’s biggest model improvements to date. They’re already making significant strides in terms of creativity, nuance, and consistency. And they’ve also (finally!) started rolling out access to their web application to everyone. At the rate they’re going, sky’s the limit for Midjourney.
For now, all we can do is wait. While you’re at it, might I suggest some of our favorite Midjourney articles such as this one? Have fun!
