There's been a lot of buzz about the best AI image generation tools lately, but not necessarily for the right reasons. Elon Musk's Grok-2 AI has stirred up a lot of controversy over its lack of guardrails. Google is perhaps hoping to shift the discussion in a more positive direction. The company has expanded access to its Imagen 3 generation tool to everyone in the US, significantly expanding the number of people who can use AI tools.
An accompanying research paper from Google describes Imagen 3 as “a latent viral model that generates high-quality images from text prompts,” adding that “Imagen 3 outperforms other state-of-the-art models at the time of evaluation.” The model's ability to easily create images based on user-entered text and has certain restrictions on offensive or illegal content makes it a rival to popular alternatives such as Midjourney.
The tool was originally announced at the Google I/O event in May 2024, but was limited to select users of Google's Vertex AI at the time. Now anyone in the US can try it out, suggesting that Google is pretty confident in the tool's ability to create reliable images and avoid some of the issues that have plagued AI image generators for years.
So far, reactions to Imagen 3 seem to be mixed. Some users have noticed that Imagen 3 is more responsive to user prompts than before, and censors words that were not blocked in Imagen 2.
For example, one Reddit user said, “It really takes extra work to get the same results as before, and even random words like 'socks' and 'water' trigger a censorship filter that's much more sensitive to innocuous words.” Another user said Imagen 3 refused to draw innocuous-sounding ideas like cyborgs and men with their arms crossed. There are positives, though, with one of the aforementioned users noting the high-quality output and “amazing textures and word recognition.”
How to try image 3
(Image courtesy of Google)
If you want to try out Imagen 3, you'll first need to be based in the United States to gain access. If you're based in the United States, head over to Google's AI Test Kitchen website and sign in with your Google account. Once you're signed in, you'll be able to start using the new image generation model.
User reports that Imagen 3 is more restrictive than its predecessor are a good example of the current state of AI image generators. These tools have generated a huge amount of controversy in recent months, and it seems like not a week goes by without an AI-generated image stirring emotions on the internet.
The latest culprit is Elon Musk's Grok-2 AI. Exclusive to premium users of X (formerly Twitter), the tool has just been updated to its second version, and uses the open source Flux to generate images. Users have found there are few limitations to what they can create, drawing pictures of Donald Trump and Kamala Harris piloting a plane on 9/11, as well as other clearly copyrighted images generated by Grok without any resistance.
With its restrictive view of image generation, Google seems to be treading cautiously with Imagen 3 for now. It remains to be seen whether that will change as the model is tweaked and improved, but now that it's accessible to a much larger audience, we may not have to wait long to see how far its limits can be pushed.
