The image generation artificial intelligence (AI) ‘Midjourney’ has been improved to the extent of generating creative images that transcend the boundaries of visual expression.

Dataconomy recently reported that Midjourney has released a new edition 5.2 of Midjourney that produces images of even higher quality.

One of the vital notable features is the ‘Zoom Out’ outpainting feature. Just as a camera could be zoomed out to suit more content on the screen, using this function expands the realm to be captured within the image to fill in additional detail. Right now, the content to be filled in by zooming out is created based on the unique image and text prompt.

The zoom out button appears after upscaling considered one of the 4 images created by the text prompt in Midjourney, and you’ll be able to select 2x, 1.5x, or Custom Zoom. Custom Zoom gives you the chance to vary the text prompt before zooming out.

We have also added a ‘panning’ feature that lets you extend a picture beyond the frame in any direction. This feature allows users to intuitively expand AI images, and use the 4 arrow buttons up, down, left, and right to create more images. Each button lets you expand the image in a single direction.

In case you need more control, you’ll be able to activate the ‘Remix’ mode by changing the prompt when using the ‘Panning’ function. Every time you click the arrows, you’ll be able to enter a recent prompt to offer you completely recent control over the image. You possibly can change the background and provides your image a totally different mood.

There’s also a “make square” option that turns non-square images into square images. There’s also a “Custom Zoom” button, a sophisticated tool that enables users to vary the text prompt and alter the aspect ratio.

Moreover, with a feature called ‘Prompt Parser’, users can now optimize their prompts for best results by identifying necessary words and removing unnecessary words.

The parser also assigns a weight to every word, allowing the user to pick from five options for probably the most appropriate prompt. Simply put, it helps transform vague ideas into prompts that higher suit the user’s intent and wishes.

Reporter Park Chan cpark@aitimes.com