This artist used machine learning to create realistic portraits of Roman emperors

Some people have spent their quarantine downtime baking sourdough bread. Others experiment with tie-dye. But others — namely Toronto-based artist Daniel Voshart — have created painstaking portraits of all 54 Roman emperors of the Principate period, which spanned from 27 BC to 285 AD.

The portraits help people visualize what the Roman emperors would have looked like when they were alive.

Included are Voshart’s best artistic guesses of the faces of emperors Augustus, Nero, Caligula, Marcus Aurelius and Claudius, among others. They don’t look particularly heroic or epic — rather, they look like regular people, with craggy foreheads, receding hairlines and bags under their eyes.

To make the portraits, Voshart used a design software called Artbreeder, which relies on a kind of artificial intelligence called generative adversarial networks (GANs).

What did I do during quarantine? 54 Machine-learning assisted portraits of the Roman Empire.

For this project, I transformed or restored (cracks, noses, ears etc.) over 800 images.https://t.co/yQWWIgfvjJ

— Dan Voshart 𓀡 𓀒 𓀓 𓀢 (@dvoshart) July 24, 2020

Voshart starts by feeding the GANs hundreds of images of the emperors collected from ancient sculpted busts, coins and statues. Then he gets a composite image, which he tweaks in Photoshop. To choose characteristics such as hair color and eye color, Voshart researches the emperors’ backgrounds and lineages.

“It was a bit of a challenge,” he says. “About a quarter of the project was doing research, trying to figure out if there’s something written about their appearance.”

He also needed to find good images to feed the GANs.

“Another quarter of the research was finding the bust, finding when it was carved… because a lot of these busts are recarvings or carved hundreds of years later,” he says.

In a statement posted on Medium, he writes: “My goal was not to romanticize emperors or make them seem heroic. In choosing bust/sculptures, my approach was to favor the bust that was made when the emperor was alive. Otherwise, I favored the bust made with the greatest craftsmanship and where the emperor was stereotypically uglier — my pet theory being that artists were likely trying to flatter their subjects.”

Voshart is not a Rome expert. His background is in architecture and design, and by day he works in the art department of the TV show “Star Trek: Discovery,” where he designs virtual reality walkthroughs of the sets before they were built.

But when the coronavirus pandemic hit, Voshart was furloughed. He used the extra time on his hands to learn how to use the Artbreeder software. Eventually, he joined a Reddit thread on colorized statues where people were posting realistic-looking images they’d created on Artbreeder using photos of Roman busts. Voshart gave it a try — and went into excruciating detail with his research.

Voshart says he made some mistakes along the way. For example, Voshart initially based his portrait of Caligula, a notoriously sadistic emperor, on a beautifully preserved bust in the Metropolitan Museum of Art. But the bust was too perfect-looking, Voshart says.

“Multiple people told me he was disfigured, and another bust was more accurate,” he says.

So, for the second iteration of the portrait, Voshart favored a different bust where one eye was lower than the other.

“People have been telling me my first depiction of Caligula was hot,” he says. “Now, no one’s telling me that.”

Voshart says people who see his portraits on Twitter and Reddit often approach them like they’d approaching Tinder profiles.

“I get maybe a few too many comments, like ‘such-and-such is hot.’ But a lot of these emperors are such awful people!”

Daniel Voshart, Toronto-based artist

“I get maybe a few too many comments, like ‘such-and-such is hot.’ But a lot of these emperors are such awful people!” Voshart says.

Voshart keeps a list on his computer of all the funny comparisons people have made to present-day celebrities and public figures.

“I’ve heard Nero looks like a football player. Augustus looks like Daniel Craig…my early depiction of Marcus Aurelius looks like the Dude from ‘The Big Lebowski.’”

But the No. 1 comment? “Augustus looks like Putin.”

No one knows for sure whether Augustus actually looked like Vladimir Putin in real life. In truth, the busts themselves aren’t all that trustworthy. In ancient Rome, sculptures often served an ideological purpose: Their subjects’ traits were supposed to communicate where rulers were from, to whom they were related or what they believed in — and not necessarily what they looked like.

Voshart says his portraits are speculative.

“It’s definitely an artistic interpretation,” he says. “I’m sure if you time-traveled, you’d be very angry at me.”

Published at Tue, 08 Sep 2020 18:56:15 +0000

Europe’s First Artificial Intelligence Space Missions

Europe’s ɸ-sat-1 launch onboard Vega rocketESA-European Space Agency

Europe’s first Artificial Intelligence Earth observation mission, ɸ-Sat-1, was successfully launched. ESA’s ɸ-Sat-2 is now underway.

Following ESA‘s successful launch of its first Artificial Intelligence (AI) Earth observation mission ɸ-Sat-1 (pronounced Phi-Sat-1), the European Space Agency (ESA) follows-up with the next innovative state-of-the-art technology: ɸ-Sat-2 (Phi-Sat-2).

In another historical event to advance European space missions, on September 3, 2020, ESA launched its ɸ-Sat-1, an enhancement of the Federated Satellite Systems mission on-board a Vega rocket from Europe’s spaceport in Kourou, French Guiana.

ɸ-Sat-1 mission is to demonstrate how satellite data, coupled with advanced onboard digital technologies, can bring benefits to businesses, industry, and science.

Europe’s first Artificial Intelligence in space’s mission: AI cloud detection, atmospheric and space science observations

The main task of the AI chip on ɸ-Sat-1 is to comb through huge sets of images and filter out the ones of low-quality due to cloud coverage.

The main advantage of processing large amounts of data on-board is that it makes the delivery to Earth more efficient as the AI has already removed the cloudy images.

The AI cloud detection experiment is going to validate the performance of the on-board inference engine based on a machine learning algorithm for cloud detection.

ɸ-Sat-1 AI

ESA has been working with various partners to demonstrate the potential of Artificial Intelligence and its applications in space to develop ɸ-Sat and to enhance the FSSCat mission.

The hyperspectral camera on one of the two CubeSats is going to collect a huge number of images of planet Earth. Some of these images will have to be discarded due to cloud coverage.

As a way of saving time and human effort back on Earth, the ɸ-Sat Artificial Intelligence chip is going to identify the unsuitable images and filter them out, only sending the images which contain usable data.

“The great interest and learning experience we gathered with ɸ-Sat-1 encouraged us to continue with ɸ-Sat-2. The positive feedback we received after evaluating 16 great mission concepts for ɸ-Sat-2 gives us the signal to continue preparing AI technology and issuing more ɸ-Sat calls in the coming years, said Josep Rosello, Head of Technology Coordination and Frequency Management Section at ESA.

“ɸ-Sat-2 is the next step in ESA’s drive for continuous innovation in Earth observation, which was initiated with the creation of the ɸ-Lab in 2017, and the launch of ɸ-Sat-1 on September 3 this year. ɸ-Sat-2 will further push the frontier of technology and open up new opportunities for the space and data analytics industry,” added ESA’s Director for Earth Observation, Josef Aschbacher.

ɸ-Sat-2 AI

The ɸ-Sat-2 mission will further demonstrate the capabilities of Artificial Intelligence (AI) technology for Earth observations following up from ɸ-Sat-1. The use of AI is going to lead to novel ways of collecting, distributing, and analyzing data collected in space about planet Earth.

Now ESA has announced that a panel of ESA experts has selected the mission proposed by a consortium as the winning idea. The consortium is led by Open Cosmos.

The consortium includes six different European countries and includes CGI, Ubotica, Simera CH Innovative, CEiiA, GEO-K, and KP Labs as partners and sub-contractors. If negotiations are successful, the ɸ-sat-2 AI satellite will be ready to start its mission just 16 months after the agreement.

The new ɸ-Sat-2 mission has been planned to address a wide range of Artificial Intelligence applications which include transforming a satellite image into a street map, cloud detection in order to reduce the huge volume of data that needs to be downloaded to the ground base, autonomous detection and classification of maritime vessels, forest monitoring and anomaly detection.

Flying in a Sun-synchronous orbit, the payload will include the AI processor Intel Movidius Myriad 2 from Ubotica. The AI processor was already adopted on the ɸ-Sat-1 mission.

Expressing his excitement about the successful launch of the ɸ-Sat-1, Massimiliano Pastera, ɸ-Sat-1 and ɸ-Sat-2 Officer at ESA said that the launch of ɸ-Sat-1 will allow them “to understand the use of AI for cloud detection, and ɸ-Sat-2 will represent a flying platform and give us the opportunity to experiment with multiple applications, as well as verify the enabling capabilities of on-board AI for Earth observation.”

Related Articles:

Published at Tue, 08 Sep 2020 18:00:00 +0000