The latest version of the popular Artificial Intelligence generator is here. Midjourney v5 can produce stunningly realistic visuals. The v5 model uses state-of-the-art tools and a new neural architecture to generate styles and designs.
It’s taken five months to complete the updated edition of Midjourney, and yes, you can now draw hands! The team announced the news on the project’s official Discord channel.
Midjourney v5: What’s new?
According to the announcement, these are the changes for version 5:
- Greater stylistic range and more responsiveness to different commands.
- Higher image quality (double the resolution) and improved dynamic range.
- More detailed images. Details are more accurate and there is less unwanted text.
- Improved performance with image stimuli.
- Support for seamless mosaic plot (experimental).
- Support for aspect ratios greater than 2:1 (experimental).
- Support for weighting image stimuli against text stimuli.
What does it all mean?
- Style: The stylistic range of Midjourney v5 is wider and better understands user requests. In recent years, the importance of mentioning the name of the artist or movement has grown. If no style is specified, the neural network will use the default style to create a photorealistic output.
- Quality: There is now twice the resolution and a wider dynamic range. High quality images can now be created instantly. In version 4, the request was processed while the user waited for each image to be enlarged by hand. From now on, clicking the Enlarge button will immediately load a larger version of the image.
- Detail: When it comes to small details, the neural network excels. Thus, it draws five fingers on each hand and does not add a hundred teeth. The developers also claim to have “drastically” improved facial detail.
This new edition of Midjourney will allow users to create photorealistic visuals that may confound some people’s ability to distinguish reality from simulation. We have come a long way in just a few months with Midjourney v4, and can now find images with a greater level of graphical detail than we could have anticipated.
The creators warn that two-word prompts may not be as effective as before and suggest creating longer and more specific stimuli. The beta development phase of the fifth generation is still in progress. The developers warn that the neural network may undergo significant changes before the final release. The generation, however, is already available to people.
How to use Midjourney v5?
Currently, the default version of the Midjourney neural network is v4. You have two options to activate the fifth:
- Simply type /settings in Discord, choose the MJ option and press Enter to apply the changes.