July 23, 2019 weblog
Mugshots evoke mood of gallery, grapes and goblets
Call them crazy but there is a cross section of human life who don't want to be turning up in social chat groups looking like cuties with red bunny noses or bunny ears. Like. Ever.
Call them vain but that cross section would want to see their faces evoke personalities that look pensive and well art-museum-ish. For the latter, a vanity-soothing solution awaits.
That is the invitation to try out what can happen when your own photo is sent over to AI Portraits Ars researchers. Their artificial intelligence tool can turn your headshot into something that looks like a great-art painting.
Built by researchers at the MIT-IBM Watson AI Lab, AI Portraits Ars goes to work to make your mugshot and it was trained on 45,000 portraits, from styles ranging from the Early Renaissance to the Contemporary period, said TNW.
"AI Portraits Ars is able to paint portraits in real time at 4k resolution," said the team.. "You will find yourself in front of a mirror and feel thousands Rembrandt, Caravaggio, Titian portraying you moment after moment."
You do not get to choose which style you want, though. The AI then "decides" which format it will take, not you, "Certain elements within any selfie may prompt the algorithm to use a specific style," said James Vincent in The Verge.
The team behind the tool said they trained AI Portraits Ars using their Generative Adversarial Network (GAN) on 45,000 portrait images. "The GAN is composed of two neural networks," they added, "one learns to recognize portraits of people (Discriminator), and the other learns to generate them (Generator)."
(Karen Hao back in January reflected on GANs in MIT Technology Review: "GANs, or generative adversarial networks, are the social-media starlet of AI algorithms. They are responsible for creating the first AI painting ever sold at an art auction...They work by pitting two neural networks against each other to create realistic outputs based on what they are fed.")
Napier Lopez, TNW: "The images might not look exactly like their sources – I couldn't render a portrait that clearly looked like me – but the end result could have fooled someone into thinking they were real paintings of a lookalike at a glance."
No doubt serious art followers would want to make the point that such tools are not to be confused as actually matching the skills and creativity of original artists, but nonetheless this exercise in pseudo-portraiture is quite interesting. And, its creators note that it can be instructive, too. At least for them. They explored their before-after process and the history of art.
Case in point: Don't expect lots of grins and teeth.
"Portrait masters rarely paint smiling people because smiles and laughter were commonly associated with a more comic aspect of genre painting, and because the display of such an overt expression as smiling can seem to distort the face of the sitter. This inability of artificial intelligence to reproduce our smiles is teaching us something about the history of art."
© 2019 Science X Network