Google Rolls Out Beta Version Of ‘Imagen’ AI Art Generator To Mobile Users

Google is finally ready to share what it has been cooking up in its AI Test Kitchen app by serving up its AI generator, Imagen, to users.
 
Imagen was first announced earlier this year when the company released a proof-of-concept paper and website for people to view its photorealistic capabilities.

And after this year has seen countless different AI imaging machines pop up on the internet, it seems like Google is ready to jump on the bandwagon.

However, instead of being able to imagine a photo of teddy bear swimming in the Olympics like in the picture above, only two functions will be available on the app. To unlock them, you will have to fill out a short form before being put on the waitlist. Once you’re in, you will find ‘City Dreamer’ and ‘Wobbler’.
 
‘City Dreamer’ allows users to describe a town or location and have the model build it for them. At the same time, ‘Wobbler’ lets you dress up and create a little monster that will wobble around when poked. 
 
While this might seem simple compared to OpenAI and Stable Diffusion, there’s a reason why Google is only releasing limited capabilities right now.
 
Before the technology is ready for a complete release, the company is primarily attempting to gather community feedback to fine-tune the tools. In addition, Google has had a front-row seat to the conversation surrounding AI art. In particular, the inherent bias that regrettably seeps into such models, how artists should be attributed, and whether this detracts from actual creators and their work.
 
As such, Google is using its AI Test Kitchen app (available on the Google Play and iOS App Stores) as an experiment to see how to develop the technology further and how to weed out all the problems that come with AI art.

[via Artnet News and TechGoing, images via Imagen]