How Accurate Are “Photorealistic” Portraits of Roman Emperors?

In 2020, the digital artist Daniel Voshart used a neural net called Artbreeder to create a series of “photorealistic” portraits of fifty-four Roman emperors spanning from Augustus (ruled 27 BCE – 14 CE) to Carinus (who died in 285 CE). As the term “photorealistic” suggests, his portraits look almost like photographs. Unfortunately, a lot of people do not realize that these portraits are modern artistic impressions, not scientific recreations of what the Roman emperors really historically looked like.

I’ve seen many people over the past year cite Voshart’s portraits and others like them as though they were authoritative, scientific recreations of what the Roman emperors really looked like. In particular, I’ve noticed a worrying number of white supremacists trying to cite these kinds of portraits as “evidence” that the ancient Romans were all white. In this article I would like to discuss why Voshart’s portraits—and others like them—should be taken with several grains of salt.

Instances of people taking Voshart’s portraits as more scientific than they really are

When Voshart first began publishing his portraits last year, lots of media outlets touted them using over-the-top headlines that emphasized his use of technology. For instance:

• An article for The Verge published on 21 August 2020 bears the headline: “How a designer used AI and Photoshop to bring ancient Roman emperors back to life.”

• An article for LiveScience published on 28 September 2020 has a headline that reads: “AI ‘resurrects’ 54 Roman emperors, in stunningly lifelike images.”

• An article for Popular Mechanics published on 5 October 2020 has an even more dramatic headline that reads: “The Real Faces of 54 Roman Emperors.” The subheading declares: “For the first time, cutting-edge tech reveals what these famous rulers actually looked like,” which is just blatantly wrong and ridiculous.

These headlines, with their emphasis on Voshart’s use of AI, have contributed to a false sense that his portraits are “scientific” and therefore automatically historically accurate. Sarah E. Bond and Nyasha Junior, however, point out in their article “How Racial Bias in Tech Has Developed the ‘New Jim Code,’” published in Hyperallergic on 8 October 2020, that this is far from the case. Voshart may have used AI to aid in the creation of his portraits, but he was still the one making them and the portraits are ultimately the result of his own, human decisions.

Unfortunately, as Bond and Junior also note in their article, Voshart’s portraits and other “photorealistic” portraits of Roman emperors have become widely admired among white supremacists, who have tried to use them as “evidence” that the Romans were racially white. For instance, on 27 September 2020, a digital artist using the Twitter handle “@ artistfuly” posted four pictures she made of the Roman poets Horace, Vergil, Ovid, and Catullus. Her pictures happened to portray the poets with deeply tanned, almost brown skin. As a result, tons of racist white people were outraged. They swarmed in to harass and belittle her, insisting that she portrayed the Roman poets as too dark and accusing her of “blackwashing.”

On 28 September—the same day that the LiveScience article I mentioned earlier came out—a person using the Twitter handle “@ Apocalypse_56” posted Voshart’s portrait of the Roman emperor Marcus Claudius Tacitus (ruled 275 – 276 CE), treating it as though it were an actual photograph of the emperor himself that somehow definitively proved that he and all other Romans were white.

Meanwhile, on Quora, there is a space for Neo-Nazis that calls itself “White Traditionalism.” The space’s URL includes the phrase “white genocide,” which is an allusion to a Neo-Nazi conspiracy theory, which falsely claims that some sinister, powerful people (who are usually stated to be Jewish people, leftists, or “globalists”) are using their malevolent influence to eradicate the white race and replace white people with people of color.

The space’s official tagline is “Embrace Traditionalism, destroy degeneracy.” Unsurprisingly, it includes regular posts about Neo-Nazi topics such as the supposed need for the creation of a so-called “white ethnostate” where white people will be able to live free from the existence of people of color.

One of the most recent posts in the space as of the time I am writing this is a screenshot of a Twitter exchange about modern photorealistic portraits of the Roman emperors Augustus, Tiberius, Hadrian, and Aurelian. I’m not sure if the portraits used in the post are actually by Voshart or by someone else, but they are definitely similar to his style. A reply beneath the initial tweet in the screenshot asks: “Why did you make them all SOO white?” A reply to this tweet reads: “Because life is not Netflix.”

This reply, which is presumably supposed to be funny or something, is almost certainly an allusion to the BBC One miniseries Troy: Fall of a City, which was distributed internationally on Netflix. The show sparked widespread outrage among racist white people on the internet who were absolutely infuriated and appalled over the fact that, in the show, Achilles is portrayed by a Black actor. I wrote a whole article about the controversy back in June 2020.

What is especially interesting is that many of the people commenting underneath the post on Quora seem to be under the impression that the creator of the portraits somehow used technology to scientifically uncover how the Roman emperors looked in historical reality. For instance, a commenter named Betzalel Schwarz declares: “Technology is astounding.”

Another commenter named Drew Fowler comments that he thinks the accuracy of the portraits is “nearly 100% on the ball”:

The problem is that it is simply not possible to scientifically reconstruct exactly what anyone from two thousand years ago looked like.

One of the best cases of an attempt to recreate an ancient historical figure’s appearance is King Philippos II of Makedonia. Philippos’s actual skull and multiple ancient sculptural portrait heads of him have survived, meaning there is a lot of solid evidence to work with. Nonetheless, even when modern forensic facial reconstruction experts try to recreate Philippos’s appearance, they are still relying on a lot of guesswork, especially when it comes to his skin and hair.

When it comes to Roman emperors, we don’t have their skulls. In the very best cases, all we have are portrait busts and heads that are usually heavily idealized. In the worst cases, all we have are tiny portraits on coins that give us no more than a very general impression of their features.

ABOVE: Photograph from Reddit of a forensic facial reconstruction of what King Philippos II of Makedonia might have looked like, based on his skull and surviving portraits of him

Skin color, hair color, and eye color

Let’s start off by talking about skin, hair, and eye color, since these are the issues that the white supremacists who keep sharing photorealistic portraits of Roman emperors are most obsessed with. All ancient Roman marble statues were originally painted, but, over the centuries, the original pigments have deteriorated so that, in most cases, they are no longer visible to the naked eye.

It is often possible to recover information about the pigments that were used on the statues by examining them extremely up close under an ultraviolet light. Researchers such as the German archaeologist Vinzenz Brinkmann have done this for many ancient Greek and Roman statues.

Notably, a color reconstruction of a portrait head of the emperor Gaius Caesar Augustus Germanicus (ruled 37 – 41 CE), better known today as “Caligula,” has been displayed as part of Brinkmann’s famous exhibit of ancient Greek and Roman polychromy, titled “Gods in Color.” Unfortunately, most surviving Roman statues have not yet been examined in this manner, meaning that, in most cases, we do not know what their original pigments were.

Ancient biographers sometimes record information about certain emperors’ skin, hair, and eye colors, but the words they use in Latin are often imprecise and have varying interpretations. For instance, the Roman biographer Gaius Suetonius Tranquillus (lived c. 69 – after c. 122 CE) records in his Life of Nero 51 that Nero’s hair was subflavus, which is usually translated as “blond.” This could, however, refer to any shade of hair ranging from slightly blondish brown to almost white on one axis and from golden to a reddish strawberry blond on another axis.

The reliability of these descriptions is also often in dispute. Roman biographers frequently wrote their descriptions of Roman emperors to incorporate physiognomic clues to the emperors’ personalities. For instance, as Rebecca Futo Kennedy, an associate professor of classics at Denison University, discusses in this thread on Twitter posted on 4 August 2021, the Romans associated paleness with weakness, cowardice, and effeminacy, so, when Suetonius describes emperors like Tiberius and Caligula as pale, he may not be accurately describing how they really looked and instead be trying to portray them as weak.

And these descriptions only even apply to some emperors. For many Roman emperors, especially those who reigned only briefly, we don’t have any descriptions of their skin, hair, or eye colors at all. This means that, practically speaking, much of the time, when artists like Voshart create their portraits of Roman emperors, they are really forced to guess what colors of skin, hair, and eyes the emperors in question had based on often vague, scanty, or nonexistent information.

ABOVE: Photograph from Wikimedia Commons showing an ancient Roman portrait head of Caligula (left) and a reconstruction of what the original colors of it might have looked like (right), both displayed as part of Vinzenz Brinkmann’s exhibit “Gods in Color”

Most people tend to assume that the ancient Romans were all what most present-day Americans would consider white. As I discuss in this article I wrote in September 2020, however, this is not an accurate assumption. The Roman Empire included southern Europe, but also all of North Africa and large swathes of the Middle East. Most people who lived in the Roman Empire were probably not what most twenty-first-century Americans would consider white. Indeed, there were certainly many Roman emperors who were not what most Americans today would consider white.

Notably, one of the few Roman emperors of whom there is a surviving color portrait is the emperor Septimius Severus (ruled 193 – 211 CE), who was born in the city of Leptis Magna in what is now Libya. His mother was of Italian ancestry, but his father was of Punic and possibly Amazigh ancestry. An imperial family portrait that was painted sometime around 199 CE, depicting him with his Syrian wife Julia Domna and their two sons Geta and Caracalla, known as the “Severan Tondo,” has survived.

The portrait clearly shows Septimius Severus with brown skin that is much darker than the skin of any of the other members of his family. This may be partly due to ancient Roman artistic conventions of portraying gender, since Roman artists frequently depict men as dark skinned and women as pale, since men were expected to work outside and become tanned while women were expected to remain indoors and remain pale.

Nonetheless, in the Severan Tondo in particular, Septimius Severus’s dark skin seems to be more than a matter of mere gender distinction and it probably does reflect what he really looked like, at least to some extent.

ABOVE: Photograph from Wikimedia Commons of the Severan Tondo, a family portrait of the Libyan emperor Septimius Severus with his Syrian wife Julia Domna and their sons Geta and Caracalla that was painted in around 199 CE or thereabouts

Despite the fact that white supremacists seem to love Daniel Voshart’s portraits, Voshart himself is clearly not a white supremacist. In fact, I think that, in general, he has actually done a remarkably good job of navigating the racial issue in his portraits. While white supremacists love to cherry-pick portraits he has done of emperors who appear to be white, his portraits—especially in their most recent versions—actually show a fairly wide range of skin colors.

It is also notable that Voshart has intentionally revised his portraits to more accurately reflect the color diversity of Roman emperors. For instance, when Voshart first published his portrait of Septimius Severus, it depicted his skin as a relatively light shade of brown, close to merely tan or even white.

ABOVE: Image from Voshart’s webpage showing his original portrait of Septimius Severus, showing him with very light brown skin

Voshart, however, later went back and revised his portrait to make Septimius Severus’s skin a darker shade of brown that is much closer to what is shown in the Severan Tondo. Voshart simultaneously also revised his portraits of Septimius Severus’s sons Geta and Caracalla to give them slightly darker complexions that are more similar to their father’s complexion.

ABOVE: Image from Voshart’s webpage showing his revised portrait of Septimius Severus, with a darker skin tone that more closely matches the skin tone depicted in the Severan Tondo

Augustus and the idealization of Roman portraits

Even if we ignore the whole matter of assumptions concerning skin color, hair color, and eye color, there are still many other problems for the accuracy of Voshart’s portraits and others like them. One particularly obvious problem is that these portraits are inevitably based on ancient Roman portraits, which were never supposed to accurately portray their subjects with photographic accuracy and precision.

Ancient Roman portraitists certainly wanted to create portraits that resembled the people they portrayed, but they freely downplayed, removed, exaggerated, and sometimes even outright invented certain features based on how the people they were portraying wanted to be portrayed.

For instance, there are literally hundreds of surviving ancient portrait busts and heads of Octavian (later known as Augustus), who was the first person to adopt the title of Princeps Civitatis (meaning “First Citizen”) and is generally considered to have been the first emperor of the Roman Principate. The surviving portraits from when Octavian was a young man appear mostly on coins. These generally depict him as a small, somewhat disheveled youth. These portraits may be meant to showcase his youthful precociousness.

ABOVE: Photograph from Wikimedia Commons of a gold aureus minted in 41 BCE, depicting Marcus Antonius on the obverse and the young Octavian on the reverse

All Augustus’s later portraits from the time when he was mature and secure in his power show him as an ideally handsome man who looks to be in his thirties or forties. These portraits are probably at least somewhat reflective of what Augustus really looked like during the early decades of his reign, but they are certainly what we might think of as heavily “airbrushed” or “Photoshopped” and they are clearly meant to portray him as an ideal, powerful ruler.

Furthermore, Augustus ruled for many decades. By the time of his death in 14 CE, he was seventy-five years old. There’s no way he still looked the same way he did when he was younger, but yet all his portraits continue to portray him as such. There are no surviving portraits of Augustus that show him as he really looked in his old age.

ABOVE: Photograph from Wikimedia Commons of a portrait bust depicting the emperor Augustus as a young man wearing the civic crown, contemporary to his reign

Voshart’s portrait of Augustus attempts to account for this problem by artificially aging him from his surviving portraits to show what he might have looked like when he was in his fifties or thereabouts. The result is probably not too far off from what Augustus really looked like, but only if we assume that the surviving ancient portraits of Augustus are at least somewhat accurate and the digital aging process mirrors the actual aging process that played out in historical reality.

ABOVE: Image from Daniel Voshart’s webpage showing his portrait of Augustus along with the main ancient portraits on which he based it

Nero and the complication of modern restorations to ancient portraits

Another problem with using portrait busts of Roman emperors to decide what they really looked like is that, in many cases, these busts have been heavily restored. For instance, the most famous bust of the emperor Nero (ruled 54 – 68 CE) is probably the one that is currently held in the Musei Capitolini in Rome.

Nonetheless, as the art historian Dr. Kira Jones discusses in this thread she wrote on Twitter in February 2019 in response to an earlier attempt by a different artist to create a photorealistic portrait of Nero based on this bust, most of what you see is actually a seventeenth-century restoration.

The only part of the bust that is actually ancient at all is a piece that makes up the nose, the eyes, the forehead, and the front fringe of the hair. Everything else dates to the seventeenth century and was probably created by a restorer based on Nero’s coin portraits. If you look closely at a photograph of the bust, you can see that the part that is genuinely ancient is a slightly different color from the rest of the bust.

ABOVE: Photograph from Wikimedia Commons of the famous bust of the emperor Nero that is currently held in the Musei Capitolini

For better or worse, this hasn’t stopped artists from relying on the bust from the Musei Capitolini for their depictions of Nero. Voshart’s portrait of Nero is based primarily on the bust from the Musei Capitolini—even though most of the bust is not actually ancient at all.

Even more striking is that fact that one of the portraits of Nero that Voshart relies on is an entirely modern painting in the style of the Flemish painter Peter Paul Rubens (lived 1577 – 1640 CE). Thus it seems that Voshart’s sources for Nero are more seventeenth-century western European than ancient Roman.

ABOVE: Image from Daniel Voshart’s webpage showing how his portrait of Nero relies on the heavily restored bust in the Musei Capitolini, as well as a completely modern painting of Nero in the style of Peter Paul Rubens

Emperors without portraits other than coins

Augustus and Nero are relatively easy emperors to create reasonably accurate portraits of, since there are many surviving ancient sculptural portraits of them. There are, however, many Roman emperors who have no identifiable sculptural portraits whatsoever whose appearances are known solely from tiny portraits on coins.

One of the first emperors shown on Voshart’s chart of whom there are no identifiable surviving portraits other than on coins is the emperor Vitellius, who ruled briefly from 19 April to 20 December 69 CE.

ABOVE: Photograph from Wikimedia Commons of a coin depicting the Roman emperor Vitellius

Early modern collectors of portrait busts of Roman emperors were so desperate to find a bust that they could say represented Vitellius that they identified an anonymous portrait head of a somewhat overweight man dating to the reign of the emperor Hadrian that is now held in the Grimani Collection in Venice as a portrait of Vitellius, based almost solely on the fact that Vitellius is reported to have been rather overweight. The head quickly became so iconic that it spawned dozens of copies and imitations and became the basis for nearly all portrayals of Vitellius for centuries.

Modern art historians, however, now recognize that the head that was identified for so long as a portrait of Vitellius is, in fact, most likely someone else, since it has some noticeable differences from coin portraits of Vitellius. (For instance, the hairline is much more receded, the hairstyle is quite a bit different, and the face is somewhat more elongated.) The head also does not reflect the artistic styles of Vitellius’s time, meaning it was probably modeled many decades after Vitellius’s death.

ABOVE: Photograph from Wikimedia Commons of a bust of the Pseudo-Vitellius type

Nonetheless, Voshart’s portrait of Vitellius is explicitly based on busts of the Pseudo-Vitellius type. His portrait also relies on an entirely modern portrait of Vitellius painted by Peter Paul Rubens, based on the Pseudo-Vitellius type.

ABOVE: Image from Daniel Voshart’s webpage showing the portraits he used to design his own portrait of Vitellius

Aurelian

The problem of only having coin portraits becomes even more apparent when it comes to the emperors who ruled during the Third Century Crisis (lasted 235 – 284 CE). Most emperors during this period ruled only briefly, were busy fighting wars throughout their entire reigns, and did not have much time or incentive to commission sculptural portraits of themselves.

This applies even to some emperors who ruled during this period who are very famous. Notably, there are no confirmed sculptural portraits of the emperor Aurelian (ruled 270 – 275 CE); the only confirmed contemporary portraits of him are on his coins. These portraits are tiny and nowhere near detailed enough to create a photorealistic portrait of him.

ABOVE: Photograph from Wikimedia Commons of a gold aureus minted by Aurelian, bearing his image

There are several anonymous busts that some scholars think might be intended to represent Aurelian, because they bear some resemblances to his portraits on his coins, but all of these busts are also proposed to represent other emperors.

For instance, there is a gilded bronze portrait head of an emperor that was found buried beneath the Capitoline Temple in Brescia that is usually thought to represent Claudius II Gothicus (ruled 268 – 270 CE). Some, however, have argued that the head may, in fact, represent Aurelian, based on similarities it bears to Aurelian’s portraits on his coins.

ABOVE: Photograph from Wikimedia Commons of a gilded bronze portrait head of an emperor found in the Capitoline Temple in Brescia, usually thought to be Claudius II Gothicus

Voshart’s portrait of Aurelian is based partly on the gilded bronze head discovered underneath the Capitoline Temple at Brescia and partly on another bust that some have argued might possibly represent Aurelian. Ironically, his portrait does not even take Aurelian’s confirmed coin portraits into account.

ABOVE: Image from Daniel Voshart’s webpage showing the two busts that he based his portrait of Aurelian on, which may or may not actually be him

Ulpia Severina and the sheer extent of speculation

Things get even more complicated when it comes to Ulpia Severina, who was the wife of Aurelian and probably the only cisgender female Roman emperor to rule the entire Roman Empire alone under her own name during the Principate. Aurelian was assassinated sometime in late summer or early autumn 275 CE. Ulpia Severina appears to have briefly ruled as emperor for an interim period of maybe a few months until the Roman Senate elected Marcus Claudius Tacitus as the new emperor on 25 September 275 CE.

The surviving historical accounts of this period are extremely spotty. These sources do not mention Ulpia Severina’s reign and instead only mention that there was an interregnum between the death of Aurelian and the accession of Tacitus. Nonetheless, there is strong evidence that Ulpia Severina briefly ruled as emperor during this time, because we know that she continued to mint her own coins after Aurelian’s death, which bear her name, along with tiny, not especially detailed portraits of her. These are the only source of information about her appearance. There are no identifiable surviving sculptural portraits of her.

ABOVE: Photograph from Wikimedia Commons of a silver antoninianus depicting Ulpia Severina, the only female Roman emperor to rule under her own name during the Principate

Voshart created a “portrait” of Ulpia Severina that is based primarily on a gender-shift of the gilded bronze head discovered underneath the Capitoline Temple in Brescia that may represent either Claudius II Gothicus or Aurelian and only secondarily on the tiny portraits of her that appear on her coins.

Again, just contemplate for a second how little information we have about her physical appearance that the only way that a digital artist can create a portrait of her is by gender-shifting a bust that represents an entirely different, biologically unrelated person.

ABOVE: Image from Daniel Voshart’s webpage showing how he based his portrait of Ulpia Severina primarily on a gender-shift of a portrait head that may or may not represent her husband

Conclusion

Daniel Voshart’s photorealistic portraits of Roman emperors are worth looking at as long as you remember what they truly are: merely a fun art project that one man who is not a professional historian did while he was in quarantine during the COVID-19 pandemic. If you think that these are scientifically sound recreations of what Roman emperors really looked like historically, then you are sorely mistaken.

To his credit, Voshart has made it very clear that he is an artist and that he does not claim scientific or historical accuracy for his portraits. For instance, in a tweet he made on 29 July 2020, he says:

“I don’t claim perfection. This is more art than science. I am not a historian. I don’t speak Latin.”

Voshart has likewise been very open about the sources he has used to create his portraits, which has been tremendously helpful to me in writing this article. If he weren’t so open about his sources, it would have been much more difficult to write this.