Midjourney's move to a dedicated website is a big deal for AI art

Midjourney is considered by many to be one of the best AI image generators in terms of its range, accuracy and the stunning realism of its output. But it's had one major drawback... until now.

Midjourney runs on the Discord social media platform. This comes with its own quirks in terms of the user interface and overall user experience, complicating both the sign up process and actual use of the model and potentially putting some people off using the tool. And it also limits the new features that can be added. however, that looks set to end as Midjourney starts to move to its own dedicated website (see our round up of AI art tutorials to learn how it works).

Midjourney has launched the alpha version of its own standalone website, with access open to existing users who have generated more than 10,000 images. A switch to a different platform might not sound like a big deal. After all, rivals like OpenAI's DALL-E and Adobe's Firefly and most recently Meta's Imagine already work on dedicated sites. But the move makes one of the most impressive AI image generators in terms of results significantly easier to use, improving its UI and UX and making it possible to add more new features not possible within the limitations of the Discord platform.

Midjourney is close to the top of our pick of the best AI image generators, but we noted that the UI was one of its drawbacks (the other big one being its highly questionable approach to copyright, since Midjourney was trained on images scraped from the web without their owners' permission). The platform has succeeded in adding many new features on Discord, including things like panning, but it's reaching the point where further growth seems like it would be difficult.

Several Midjourney users with access to the new site have shared videos of how it looks and works, and it looks a lot easier to use. It has intuitive sliders, and users no longer have to learn the correct codes to use in their text prompts to give the model certain instructions. Some have reported that the site does not yet support rerolling or the faceswapping add on, but more features are likely to come.

Midjourney says the site also has improved social mechanics. Users can browse, search, and filter images more easily, and can comment, like and share them with others. Users will be able to customise their profiles, settings, and preferences.

The platform is also working on a Midjourney mobile app for iOS and Android, which will allow users to generate images with voice prompts as well as by writing text. Other AI image generators, such as Google Imagen 2 and DALL-E 3 have been catching up with Midjourney in terms of the realism of their output, the reliability of their interpretation of longer prompts and their handling of text in images, but we're expecting Midjourney to make another leap forward with the imminent release of version v6, which is expected before Christmas.

For more on the latest developments in AI image generation, see the news on the release of and the Instagram AI backdrop editor.