Google’s Arts & Culture app has just introduced Art Selfie 2, a tool to insert yourself into famous art styles using generative AI. This tool is a sequel to the Art Selfie feature introduced to the app back in 2018. The Arts & Culture app is a place for a dedicated community to digitize and preserve art and artifacts from around the world. While this is an excellent learning resource, Google is always introducing newer and more fun tools to help promote knowledge of art. Art Selfie 2 aims to do just that, using the allure of AI magic.
What is Art Selfie 2 all about?
The original Art Selfie feature encouraged users to submit selfies which were then compared to famous works of art. Any artworks with matching similarities were shown to the user, alongside facts and trivia. Art Selfie 2 takes this further, allowing you to take a selfie and insert it into a famous art style. The app is boasting 25 different styles of art. These include Da Vinci’s Ermine, The Great Wave, and Georgian Era amongst others.
How it works is, you take a picture of yourself, and the AI blends it into the art style of your choosing. Google says the AI will refrain from altering your face, though other aspects may be changed to fit the art style. For example, say you choose the art style “Medieval Knight”. This will not only place you in front of a castle but also cover your head with a metal helmet. “Astronaut in Space”, on the other hand, will replace your clothing with a spacesuit.
How does Art Selfie 2 work?
For those who haven’t been keeping up with advancements in generative AI, this probably seems like Sci-Fi. In short, generative AI is trained upon hundreds of thousands, if not millions of inputs. While Large Language Model generative AI is trained on posts, comments, and articles, visual generative AI is trained on art and real-life photographs. DALL-E 3 and Midjourney are two of the most popular image-generation AI models available today. And both have been trained on massive amounts of visual data.
This training ensures that the AI model is intimately acquainted with the tiniest of details. Furthermore, continuous improvements are being made to these models every day. So when you upload a selfie and request a specific art style, it’s a piece of cake for the AI. It can immediately and almost flawlessly recognize exactly what changes need to be made. Also, it is able to alter the image as well; blending unchanged bits with the new additions perfectly. And so, within mere seconds, Art Selfie 2 churns out a picture of you dressed as a Roman gladiator. Best of all, it also provides facts about the chosen art style. So users learn as they laugh at their goofy selfies.
Google has promised “many more” art styles making their way to the app soon. And as generative AI continues to improve, we’re sure to see even more impressive tools being made available to the public.
2024-01-25 15:06:07