People with a Google account can instantly try out Google’s latest AI-powered image generator.
Google’s latest AI image creation tool, called Imagen 3, is now available to the public in the United States, and only a free Google account is needed to use it. According to VentureBeat, Google quietly released this model this week and published its research in a paper on Tuesday. This release comes two months after the company first introduced the model at Google I/O in May.
In a post on Hugging Face, a machine learning platform, Google researchers announced, “We present Imagen 3, a latent diffusion model that creates high-quality images from text prompts. We outline our evaluations of quality and responsibility, noting that Imagen 3 outperforms other leading models based on our assessments. We also address safety and representation concerns and the steps we’ve taken to reduce potential harm from our models.”
From my initial tests, the model appears to be quite reliable. When asked to generate photorealistic images, the results are generally high-quality, with a convincing level of realism that might deceive viewers at first glance. I was especially impressed with the quality of images when I requested a retro 35mm film effect. Imagen 3 also emphasizes key elements of the prompt that influenced the result, letting you adjust these if you’re not satisfied with the final image.
Despite its impressive results, Imagen 3’s outputs still exhibit some signs of being AI-generated. In some images, hands may have an unusual number of fingers, faces can be distorted, and text may be nonsensical. However, the model did manage to accurately reproduce logos like “Coca-Cola” and “Canon,” which could potentially infringe on trademarks.
Google isn’t the only tech company unveiling a new image model this week. X has also launched a beta version of Grok, its AI chatbot, which includes an image generator with seemingly few restrictions. Users, such as Michelle Ehrhardt from Lifehacker, have already used Grok to create various content, including images of Taylor Swift wearing a MAGA hat and Pikachu holding an AK-47.
On the other hand, Imagen 3 has evident safeguards in place. When I tried to use Google’s image generator for sensitive topics, it stopped the process and kindly directed me to Imagen 3’s FAQ to explain why the prompt was unsuitable. It also avoids generating copyrighted material but can be manipulated into producing such content with specific prompts. For instance, I was able to reproduce logos and even coax trademarked characters like Mario and Pikachu, although they didn’t engage in any shootouts.
How to use Imagen 3
Google’s new AI image generation tool is available for free to any U.S. user with a Google Account. To access it, go to ImageFX in Google’s AI Test Kitchen. Just sign in with your Google Account, and you can start creating images right away.
The launch of Imagen 3, Google’s innovative AI image generation tool, has generated excitement in the tech community. This new tool allows users to produce high-quality images from text prompts, marking a significant addition to the array of AI tools available in the tech industry.
Additionally, major tech companies like Google and X are relentlessly advancing AI technology, evident in Google’s Imagen 3 and X’s new image generator. These innovations are creating new opportunities for content creation, as illustrated by the wide range of images produced with Grok.