The alpha release of Midjourney V6 will be accessible on Discord during the winter break, following nine months of development. This update, available for a brief period starting today, left a striking impression during my initial encounter. While Gizmodo previously tested prominent AI image generators, Midjourney V6 surpasses them all. The generated images exhibit an exceptional level of realism without compromising the inherent creativity that has been a hallmark of Midjourney. Unlike many AI image generators that force a choice between realism and fantasy, Midjourney V6 adeptly combines both aspects in a way I haven’t seen before.
Similar to its predecessors, accessing Midjourney V6 requires initiating in a Discord server with the Midjourney Bot. By typing “/settings” in the chat bar, users can select from a list of available Midjourney versions, with V5.2 likely being the default. For a limited time, users can switch to V6 and proceed to create prompts by typing “/imagine” followed by their desired input.
Alternatively, users can access Midjourney V6 by appending “–v 6” to the end of their prompts. Although this method may seem a bit less streamlined, it functions equally well. Developers emphasize that V6 necessitates a different prompting approach than its predecessors. Given its improved understanding, users are encouraged to be explicit in their prompts. For a more realistic outcome, appending “–style raw” at the end of the prompt is recommended instead of relying solely on descriptive language.
While the updated version exhibits a slightly slower performance compared to other Midjourney models, the generated photos are remarkably realistic. Consequently, Midjourney has enhanced its moderation systems in response to the convincing nature of the images. Developers stress a stricter enforcement of community standards to discourage inappropriate behavior and image manipulation for deceptive purposes.