Midjourney 6 Release

This version brings several improvements and new features, enhancing the overall user experience and the quality of the generated images.

One of the most significant improvements in Midjourney V6 is its enhanced understanding of natural language prompts. Users no longer need to use specific terms like "4K", "Unreal Engine", or camera model names. Instead, they can describe the image they want to create just as they would to a human and receive results close to their description.

The photorealism of the images generated by Midjourney V6 has also been improved. The images are now more coherent and detailed, making the platform even more useful for creating realistic images.

Another interesting update is the ability to generate multiple characters and images of real people like actors, politicians, and celebrities. However, due to privacy concerns, this feature might be removed or more heavily censored in the future.

Midjourney V6 can now also generate short text on images. While this feature is not super impressive right now, the developers have promised that it will soon be significantly improved.

The platform also offers better art styles. For instance, V6 can generate very high-quality and precise pixel art images. Other image styles have also been noticeably enhanced.

However, some features from version 5, like Zoom inpainting, 4X upscaling, and style tuning, are not currently available in V6. These features will be reintroduced gradually in the coming months.

Midjourney V6 is accessible via Discord, and it offers different subscription plans.

Despite these improvements, some users believe that Dall-E 3 (my favorite) is still superior in understanding and following prompts. But Midjourney V6 has functionalities that Dall-E 3 still does not, such as mixing images and creating variations of a created image.

Sources:

https://ppl-ai-file-upload.s3.amazonaws.com/web/direct-files/1357227/97167e1d-ec62-4962-aa5a-13ea476d47d8/Midjourney_6.pdf

https://youtube.com/watch?v=ELzMt2KFI9E