Discover Midjourney Alpha: Web-Based AI Image Generation Unleashed

Midjourney, one of the leading AI image generation services globally, boasting over 17.5 million users on Discord, has launched an alpha version of its website, midjourney.com. This new platform allows users to generate images directly from the site, moving away from the Discord application.

The alpha version was first discovered by active users like Nick St. Pierre (@NickFloats), who shared screenshots and videos on X. Today, Midjourney confirmed this testing phase in their Discord announcements, indicating that access is currently limited to those who have created more than 10,000 images using the service. The announcement reassured users that broader access will be rolled out in the coming month.

To check their image generation count, users can log into Midjourney on Discord and use the command “/info” in any channel or directly with the Midjourney bot. Despite having generated 8,383 images, I was able to access the Midjourney Alpha website and test its features.

Interface and Functionality

The website presents a clean, minimalistic interface with ample white space. Upon landing on the site, users find an “Explore” tab showcasing public image generations. At the top, a simple rectangular text box labeled “Imagine” allows eligible users to input their prompts.

To the right of this bar, an icon representing dials opens a visual interface for adjusting prompt settings, a significant improvement over the Discord setup, which requires users to manually enter text commands. For example, instead of typing “–ar 4:3” for a landscape image, users can now just click the “Landscape” option, automatically applying the setting without cluttering the prompt.

The left sidebar features a “Photo” icon that displays a reverse-chronological grid of all the user's past generations, grouped by date, as well as their current image generation jobs.

Competitive Edge

The Midjourney Alpha website displays similarities to Visual Electric, a newly launched rival geared toward professional designers and based on the open-source Stable Diffusion model. However, many users believe Midjourney’s image quality surpasses that of its competitors. The Alpha site also features a “Likes” tab showcasing images that users have previously favored on Discord, as well as a searchable prompt bar for exploring public images generated under specific keywords, such as “steampunk.”

Additional features include a toggle button for switching between light and dark modes and a “thumbs up” button for participating in a crowdsourced image ranking project. This element invites users to evaluate two images presented by the Midjourney team, contributing to the improvement of their AI model.

One limitation of the web interface is that generated images do not appear in the Midjourney Discord Bot chat. As a result, users who switch between platforms will find that only the web interface retains all generations from both sources.

Overall, this alpha release represents a significant step toward making Midjourney more accessible. Currently, users still need to sign in with their Discord username. Moreover, Midjourney, like other AI image generation services, faces legal scrutiny from artists over alleged copyright infringement related to its training data.

Future Developments

According to power user St. Pierre, further enhancements, including an iOS app and web folders, are on the horizon.