I didn’t take this photo in a coffee shop, but a new AI tool from Microsoft created it


That picture above? I did not eat it while writing this article in the café. I created it sitting at my home desk, using Microsoft’s new AI image generator, MAI-Image-1.

In AI jargon, the new tool is referred to as a “text-to-image form,” which means you write a prompt, and then the AI ​​tool delivers an image. The prompt I used for this image was “Create an image of two people talking and having coffee in a café.”

But I could have easily used the image below instead, since the tool gave me two images to choose from.

AI image of people in a cafe

Not real people, not real coffee.

Microsoft

Microsoft’s new AI image creator hasn’t been officially released to the market yet. You can access it by visiting Why? – A community platform where people test different AI models. LMArena also has a rating system: people compare two AI models side-by-side and vote for the model with the best AI output.

At the time of publishing this article, Microsoft’s MAI-Image-1 was ranked in the top 10 in LMArena’s rankings.


Don’t miss any of our unbiased technical content and lab reviews. Add CNET As Google’s preferred source.


In it Blog postMicrosoft said the tool is designed to avoid “repetitive or generic output.” The AI ​​team added that they are receiving feedback from creative professionals.

ai-atlas-directory-logo.png

“MAI-Image-1 excels at creating realistic images, such as lighting (e.g., bounced light, reflections), landscapes, and more,” the post said. “The combination of speed and quality means users can get their ideas on screen faster, iterate quickly, and then move their work to other tools to continue improving.”

Creating images using artificial intelligence is not new to Microsoft products. You can also create images using Microsoft Copilot or Bing Image Creator, but they are powered by OpenAI’s AI models.

Microsoft wants to develop artificial intelligence in-house. In August, The company released MAI-Voice-1 – Natural speech generator – MAI-1 preview – Text generator.

Microsoft eventually said it will bring MAI-Image-1 to Copilot and Bing Image Creator.

How to try it

Picture of wolves

The prompt to “Create an image of two coyotes playing next to a cactus” returned this image from MAI-Image-1.

Microsoft

If you want to generate images with MAI-Image-1, go to LMArena’s image generator webpage here. In the upper-right corner of the page, you’ll see a drop-down menu. “Battle” appears as the default option, but scroll to the bottom of the list and click “Live Chat.” Next, you will see another drop-down menu at the top. Scroll down until you see “mai-image1,” then select that. Then direct the AI ​​to create an image for you.

At LMArena, you can also check out other AI-powered image generators and compare the results. Instead of the live chat option, you can try a side-by-side option. You can then issue an AI image prompt, say “Show me two cats playing,” and compare the images generated by two different AI models.



Leave a Reply

Your email address will not be published. Required fields are marked *