‘Amazing’ – Midjourney update wows AI artists with camera-like feature – Ars Technica

On Thursday, Midjourney unveiled version 5.2 of its AI-powered photomontage model, which includes a new “zoom out” feature that allows a central composite image to be preserved while automatically creating a larger scene around it, simulating zoom out using a viewfinder.

is similar to External board– One of OpenAI’s DALL-E 2 imaging technologies introduced in August 2022 – Midjourney’s Zoom Out feature can take an existing AI-generated image and expand its boundaries while preserving its original subject in the center of the new image. But unlike DALL-E and Photoshop’s generative fill feature, you can’t select a custom image to enlarge it. At the moment minify v5.2 only works on images created within Midjourney, a Subscription AI image generator service.

On the Midjourney Discord server (it remains the official interface for Midjourney, though plans are underway to change this), users can experiment with minification by creating any v5.2 image (now the default) and upscaling the score. After that, special “Zoom” buttons appear below the output. You can zoom out by a factor of 1.5x, 2x, or a custom value between 1 and 2. Another button, called “Make Square,” will create texture around the current image in such a way that it creates a square aspect ratio of 1:1.

Midjourney creator David Holz announced new v5.2 features and improvements on his Discord server Thursday night. Aside from “Reduce,” the more significant additions include an aesthetic overhaul system, which promises better image quality and a stronger “Style” command that effectively affects how unrealistic an image looks. There’s also a new “High Contrast Mode”, activated by default, that increases compositional diversity between image generations. In addition, a new “/shorten” command allows users to rate prompts in an effort to trim out unnecessary words.

Despite the immediate rollout of version 5.2, Holz emphasized in his announcement that changes may occur without notice. Older versions of Midjourney are still available with the “/settings” command or the “-v 5.1” command argument on the line.

For fans of this new photomontage art form that proponents like Julie Weiland sometimes call “montage,” the changes in version 5.2 are welcome changes, with some Midjourney users calling them “surprising” And “Brainstorming,” which isn’t an unusual superlative in the hype-friendly world of AI right now. But fans will likely argue that Midjourney’s visual improvements justify the incredible awesomeness Interactions among them.

The latest update is part of a series of quality improvements since March 2022, when the model was I was born Relatively indefinite images lacking in detail. More recently, Midjourney introduced v5.0 in March and v5.1 in May this year, both of which improved realism and image detail. The introduction of the v5 mockup series allowed the creation of realistic images of Pope Francis and Donald Trump that raised concerns about deepfakes on social media.

Despite the excitement over the new features among Midjourney fans, photomontage is still controversial among some artists due to how these AI systems are trained, using millions of images scraped from the web. without Artist consultation, credit or permission. Midjourney has never officially disclosed the exact contents of its training data. Adobe is trying a more ethical path forward with Firefly, but Venture Beat recently reported that active artist consent is still marginal.

Right now, it’s hard not to appreciate the technical advances with which Midjourney opened while still wondering if there might be a more ethical path forward for this technology—one that would satisfy artists, both traditional and musicians alike.