The best AI image generators have been in the news lately, but not always for the right reasons – with Elon Musk’s Grok-2 AI causing all sorts of controversy for its lack of guardrails. Google is likely looking to steer the conversation in a more positive direction, given that the company has just secured access to its Image 3 generator for everyone in the US, vastly increasing the number of people who can use the AI tool.
from Google accompanying research document describes Imagen 3 as “a latent diffusion model that generates high-quality images from text prompts,” adding that “Imagen 3 is preferred over other state-of-the-art models at the time of evaluation.” It can be used to create images from user-entered text, with some restrictions on offensive or illegal content, making it a rival to popular alternatives such as Midjourney.
The tool originally launched in May 2024 at Google I/O, but was limited to select users of Google’s Vertex AI at the time. Now, anyone in the US can try it out if they want, suggesting that Google has much more confidence in its tool’s ability to generate reliable images and avoid some of the problems that have long plagued AI image generators.
So far, it seems that the reactions to Imagen 3 have been mixed. Several users have noted that it seems to be much more sensitive to user prompts than before, and is censoring words that may not have been blocked in Imagen 2.
For example, one user on Reddit said: “I really have to put in extra work to get what I used to get and a random word like “sock” or “water” will trigger the censorship filter which is much more sensitive to innocent words.” For anotherImagen 3 refused to draw innocent-sounding ideas, such as a cyborg or a man with his arms crossed. However, there are positives, as one of the aforementioned users noted the high quality of the output and the “amazing texture and word recognition.”
How to Try Imagen 3
If you want to try out Imagen 3, you’ll first need to live in the US to get access. If so, you can head over to Google’s AI test kitchen website and sign in with your Google account. Once that’s done, you should be able to use the new image-generation model.
User reports that Imagen 3 is more restrictive than its predecessor are a useful illustration of the current state of AI image generators. These tools have generated a lot of controversy in recent months, and it sometimes seems like a week doesn’t go by without an AI-generated image stirring up emotions across the internet.
The latest culprit is Elon Musk’s Grok-2 AI. The tool, which is limited to premium users of X (formerly Twitter), has just been updated to its second version and generates images using the open-source Flux. Users found that there were few restrictions on what could be created, leading to people drawing pictures of Donald Trump and Kamala Harris flying a plane on 9/11, or clearly copyrighted images that were generated without any pushback from Grok.
With its restrictive perspective on image generation, it seems like Google has taken a cautious approach to Imagen 3 for now. Whether that will change as the model is tweaked and refined remains to be seen, but with far more people now having access to it, we may not have to wait long to see just how far its boundaries can be pushed.