In a year absolutely dominated by the rise of artificial intelligence, it’s fitting that Midjourney has decided to cap the year with what is perhaps their most ambitious update yet. Midjourney version 6 (V6) has just released an alpha release of their latest version and it’s already behind heralded as a major step forward for the popular image generation AI model.

In particular, it looks like Midjourney V6 is finally going to crack the in-image text issue that has been hounding AI image and video generations for the better part of this formative year. Midjourney V6 also promises to bring even more photorealism to its generations and should once again lead the pack of the major AI models as one of the best text-to-image apps in the space.

Let’s take a look at Midjourney V6 and explore what its latest improvements can forecast for 2024 and beyond.

Midjourney Version 6 is Here

Developed by the Midjourney research collective and founded by David Holz, Midjourney has been one of the biggest names in AI for the better part of the past year as AI has rocketed to the forefront of pretty much all tech news. Midjourney has risen as a Discord server that, for many, has served as a gateway into the world of generative AI with a simple-to-use prompt-based structure that’s been fun and easy to play around with.

Yet, as we’ve seen in the past few months in particular, the AI wars are now in full gear and — despite Midjourney’s early lead perhaps — other new AI models have pushed ahead with so many other innovations in text-to-image, text-to-video and image-to-image.

But this newest version promises to “generate more realistic imagery” than anything that they’ve released before, according to a message on Midjourney’s Discord server from Holz, who also clarifies that this V6 version is the “third model trained from scratch” on their AI superclusters and took nine months to develop.

In-Image Text is Solved?

As we’ve covered with other AI ventures like OpenAI’s DALL-E and others (like the ambitious Ideogram launch) these issues with in-image text have hounded so many promising AI models over the past year. For what I’m sure are very technical reasons that might be difficult for us non-AI savants to understand, in-image text has been a particularly difficult nut to crack as we’ve seen results that muddle not only the spelling of words, but even the letters themselves.

However, from early samples of Midjourney V6 shared online, it looks like this latest update might have finally solved this issue. And if that’s true, the floodgates will most likely be opened for other AI models to fill in soon after as well.

How to Use Midjourney V6

As mentioned above, this latest version of Midjourney has only been released as an alpha version so far. It should be slowly rolled out into the Discord server and might only be available to select power users here at first.

That being said, there are reports that users have been able to turn V6 on by accessing the settings in the Midjourney Discord server via a direct message to the Midjourney bot. Or, it’s possible to use by manually typing in a “–v 6” after your prompts to let the bot know you want the latest model.