What Did the Roman Emperors Look Like?: See Photorealistic Portraits Created with Machine Learning

We can spend a life­time read­ing his­to­ries of ancient Rome with­out know­ing what any of its emper­ors looked like. Or rather, with­out know­ing exact­ly what they looked like: being the lead­ers of the might­i­est polit­i­cal enti­ty in the West­ern world, they had their like­ness­es stamped onto coins and carved into busts as a mat­ter of course. But such artist’s ren­der­ings inevitably come with a cer­tain degree of artis­tic license, a ten­den­cy to mold fea­tures into slight­ly more impe­r­i­al shapes. See­ing the faces of the Roman Emper­ors as we would if we were pass­ing them on the street is an expe­ri­ence made pos­si­ble only by high tech­nol­o­gy, and high tech­nol­o­gy devel­oped six­teen cen­turies after the fall of the Roman Empire at that.

“Using the neur­al-net tool Art­breed­er, Pho­to­shop and his­tor­i­cal ref­er­ences, I have cre­at­ed pho­to­re­al por­traits of Roman Emper­ors,” writes design­er Daniel Voshart. “For this project, I have trans­formed, or restored (cracks, noses, ears etc.) 800 images of busts to make the 54 emper­ors of The Prin­ci­pate (27 BC to 285 AD).”

The key tech­nol­o­gy that enables Art­breed­er to con­vinc­ing­ly blend images of faces togeth­er is what’s called a “gen­er­a­tive adver­sar­i­al net­work” (GAN). “Some call it Arti­fi­cial Intel­li­gence,” writes Voshart, “but it is more accu­rate­ly described as Machine Learn­ing.” The Verge’s James Vin­cent writes that Voshart fed in “images of emper­ors he col­lect­ed from stat­ues, coins, and paint­ings, and then tweaked the por­traits man­u­al­ly based on his­tor­i­cal descrip­tions, feed­ing them back to the GAN.”

Into the mix also went “high-res images of celebri­ties”: Daniel Craig into Augus­tus, André the Giant into Max­imi­nus Thrax (thought to have been giv­en his “a lantern jaw and moun­tain­ous frame” by a pitu­itary gland dis­or­der like that which affect­ed the colos­sal wrestler). This par­tial­ly explains why some of these uncan­ni­ly life­like emper­ors — the biggest celebri­ties of their time and place, after all — look faint­ly famil­iar. Though mod­eled as close­ly as pos­si­ble after men who real­ly lived, these exact faces (much like those in the arti­fi­cial intel­li­gence-gen­er­at­ed mod­ern pho­tographs pre­vi­ous­ly fea­tured here on Open Cul­ture) have nev­er actu­al­ly exist­ed. Still, one can imag­ine the emper­ors who inspired Voshart’s Prin­ci­pate rec­og­niz­ing them­selves in it. But what would they make of the fact that it’s also sell­ing briskly in poster form on Etsy?

Vis­it the Roman Emper­or Project here. For back­ground on this project, vis­it here.

Relat­ed Con­tent:

Five Hard­core Deaths Suf­fered By Roman Emper­ors

Play Cae­sar: Trav­el Ancient Rome with Stanford’s Inter­ac­tive Map

Rome Reborn: Take a Vir­tu­al Tour of Ancient Rome, Cir­ca 320 C.E.

The His­to­ry of Ancient Rome in 20 Quick Min­utes: A Primer Nar­rat­ed by Bri­an Cox

The His­to­ry of Rome in 179 Pod­casts

Roman Stat­ues Weren’t White; They Were Once Paint­ed in Vivid, Bright Col­ors

Based in Seoul, Col­in Mar­shall writes and broad­casts on cities, lan­guage, and cul­ture. His projects include the book The State­less City: a Walk through 21st-Cen­tu­ry Los Ange­les and the video series The City in Cin­e­ma. Fol­low him on Twit­ter at @colinmarshall, on Face­book, or on Insta­gram.

When We All Have Pocket Telephones (1923)

From Eng­land’s Dai­ly Mir­ror (Jan­u­ary 23, 1923).

Find more time­ly pre­dic­tions in the Relat­eds below.

via Neil Gaiman

Relat­ed Con­tent:

A 1947 French Film Accu­rate­ly Pre­dict­ed Our 21st-Cen­tu­ry Addic­tion to Smart­phones

Niko­la Tesla’s Pre­dic­tions for the 21st Cen­tu­ry: The Rise of Smart Phones & Wire­less, The Demise of Cof­fee, The Rule of Eugen­ics (1926/35)

In 1911, Thomas Edi­son Pre­dicts What the World Will Look Like in 2011: Smart Phones, No Pover­ty, Libraries That Fit in One Book

In 1964, Isaac Asi­mov Pre­dicts What the World Will Look Like Today: Self-Dri­ving Cars, Video Calls, Fake Meats & More

Jules Verne Accu­rate­ly Pre­dicts What the 20th Cen­tu­ry Will Look Like in His Lost Nov­el, Paris in the Twen­ti­eth Cen­tu­ry (1863)

by | Permalink | Make a Comment ( 1 ) |

A Virtual Tour of Ancient Rome, Circa 320 CE: Explore Stunning Recreations of The Forum, Colosseum and Other Monuments

If you’re a reg­u­lar read­er of this site, you’re like­ly famil­iar with the sim­u­la­tion hypoth­e­sis, the idea that con­scious expe­ri­ence is noth­ing more than a com­put­er pro­gram. This con­cept has many sci-fi impli­ca­tions, from Matrix-like sce­nar­ios to the rad­i­cal idea that every­thing in the uni­verse is soft­ware, run by incom­pre­hen­si­ble beings who might as well be gods. One of the more plau­si­ble ver­sions sug­gests that we are liv­ing in an “ances­tor sim­u­la­tion,” designed by future human soci­eties to recre­ate their past.

Pre­sum­ably, sim­u­lat­ed ances­tors would cre­ate their own ances­tor sim­u­la­tions and so on, ad infini­tum. There’s no way to know where on the con­tin­u­um we fall, but wher­ev­er it is, ances­tor sim­u­la­tions are on the way… maybe. They’re rudi­men­ta­ry at the moment, con­sist­ing of immer­sive video games and VR recre­ations of ancient cities.

Each iter­a­tion, how­ev­er, is bet­ter than the last, as we have seen in the case of Rome Reborn (or Rome Reborn®), a 3D dig­i­tal mod­el­ing project designed to recre­ate the city’s archi­tec­ture as it was in 320 CE, through expert ren­der­ings informed by archi­tec­tur­al his­to­ri­ans and “vir­tu­al archae­ol­o­gists” like Dr. Bernard Frisch­er, pro­fes­sor emer­i­tus at the Uni­ver­si­ty of Vir­ginia.

Back in a 2012 Open Cul­ture post, Matthias Rasch­er explained the sig­nif­i­cance of this year, “when Rome’s pop­u­la­tion had reached its peak (about one mil­lion) and the first Chris­t­ian church­es were being built.” His­to­ri­ans will also rec­og­nize 320 as fol­low­ing direct­ly on the heels of the Dona­tion of Con­stan­tine that gave the city to the Pope. We can tour the vir­tu­al streets of this rapid­ly chang­ing ancient city, though the bur­geon­ing pop­u­la­tion is nowhere in evi­dence. Noth­ing moves, grows, or changes in Rome Reborn. In that sense it is still like so many pre­vi­ous rep­re­sen­ta­tions of antiq­ui­ty.

Now in ver­sion 3.0, Rome Reborn began as a 3D mod­el in 2007, and was first owned by the Regents of the Uni­ver­si­ty of Cal­i­for­nia. It now oper­ates, under the aus­pices of the Uni­ver­si­ty of Vir­ginia, as a pri­vate com­pa­ny called Fly­over Zone. They have oth­er such dig­i­tal recre­ations in their prod­uct line, includ­ing “Athens Reborn®, Hadri­an’s Vil­la Reborn®, Baal­bek Reborn®, Egypt Reborn®, and His­tor­i­cal Games®.” Rome Reborn’s design­er, Dani­la Logi­nov, has released increas­ing­ly detailed pro­mos of the project over the years, and you can see these many videos here.

To ful­ly expe­ri­ence this sim­u­lat­ed Rome, you’ll need a Vir­tu­al Real­i­ty head­set. The third ver­sion of the 3D mod­el has been made pub­licly avail­able. “You can immerse your­self in the ancient city and even enter into some of its most famous build­ings while lis­ten­ing to the com­men­tary of high­ly qual­i­fied experts,” the Rome Reborn site promis­es. Famous build­ings one might explore include the Roman forum and the Basil­i­ca of Max­en­tius. It is not an expe­ri­ence based in real­ism. In some of the sim­u­la­tions “you can opt for a whirl­wind  fly­over tour of the city,” notes Meilan Sol­ly at Smith­son­ian.

This rough­ly two-hour tour is like noth­ing any ancient Roman ever expe­ri­enced. “Com­par­a­tive­ly, the two site vis­its place users in the driver’s seat,” Sol­ly writes, “afford­ing them free­dom to roam through recon­struct­ed streets and halls.” It’s not quite the stuff of a sim­u­lat­ed uni­verse just yet, but it may not be too far in the future before Rome Reborn® ful­ly lives up to its name. Learn more about ancient Rome, cir­ca 320 CE, in the videos here, and learn more about Rome Reborn at their offi­cial site.

Relat­ed Con­tent:

Explore Ancient Athens 3D, a Dig­i­tal Recon­struc­tion of the Greek City-State at the Height of Its Influ­ence

An Ani­mat­ed Recon­struc­tion of Ancient Rome: Take A 30-Minute Stroll Through the City’s Vir­tu­al­ly-Recre­at­ed Streets

French Illus­tra­tor Revives the Byzan­tine Empire with Mag­nif­i­cent­ly Detailed Draw­ings of Its Mon­u­ments & Build­ings: Hagia Sophia, Great Palace & More

Josh Jones is a writer and musi­cian based in Durham, NC. Fol­low him at @jdmagness

 

Revisit Scenes of Daily Life in Amsterdam in 1922, with Historic Footage Enhanced by Artificial Intelligence

Welkom in Ams­ter­dam… 1922.

Neur­al net­work artist Denis Shiryaev describes him­self as “an artis­tic machine-learn­ing per­son with a soul.”

For the last six months, he’s been apply­ing him­self to re-ren­der­ing doc­u­men­tary footage of city life—Belle Epoque ParisTokyo at the start of the the TaishĹŤ era, and New York City in 1911—the year of the Tri­an­gle Shirt­waist Fire.

It’s pos­si­ble you’ve seen the footage before, but nev­er so alive in feel. Shiryaev’s ren­der­ings trick mod­ern eyes with arti­fi­cial intel­li­gence, boost­ing the orig­i­nal frames-per-sec­ond rate and res­o­lu­tion, sta­bi­liz­ing and adding color—not nec­es­sar­i­ly his­tor­i­cal­ly accu­rate.

The herky-jerky bustling qual­i­ty of the black-and-white orig­i­nals is trans­formed into some­thing fuller and more flu­id, mak­ing the human sub­jects seem… well, more human.

This Trip Through the Streets of Ams­ter­dam is tru­ly a blast from the past… the antithe­sis of the social dis­tanc­ing we must cur­rent­ly prac­tice.

Mer­ry cit­i­zens jos­tle shoul­der to shoul­der, unmasked, snack­ing, danc­ing, arms slung around each oth­er… unabashed­ly curi­ous about the hand-cranked cam­era turned on them as they go about their busi­ness.

A group of women vis­it­ing out­side a shop laugh and scatter—clearly they weren’t expect­ing to be filmed in their aprons.

Young boys look­ing to steal the show push their way to the front, cut­ting capers and throw­ing mock punch­es.

Sor­ry, lads, the award for Most Mem­o­rable Per­for­mance by a Juve­nile goes to the small fel­low at the 4:10 mark. He’s not ham­ming it up at all, mere­ly tak­ing a quick puff of his cig­a­rette while run­ning along­side a crowd of men on bikes, deter­mined to keep pace with the cam­era per­son.

Numer­ous YouTube view­ers have observed with some won­der that all the peo­ple who appear, with the dis­tant excep­tion of a baby or two at the end, would be in the grave by now.

They do seem so alive.

Mod­ern eyes should also take note of the absences: no cars, no plas­tic, no cell phones…

And, of course, every­one is white. The Nether­lands’ pop­u­la­tion would not diver­si­fy racial­ly for anoth­er cou­ple of decades, begin­ning with immi­grants from Indone­sia after WWII and Suri­nam in the 50s.

With regard to that, please be fore­warned that not all of the YouTube com­ments have to do with cheeky lit­tle boys and babies who would be push­ing 100…

The footage is tak­en from the archival col­lec­tion of the EYE film­mu­se­um in Ams­ter­dam, with ambi­ent sound by Guy Jones.

See more of Denis Shiryaev’s  upscaled vin­tage footage in the links below.

Relat­ed Con­tent:

Watch Vin­tage Footage of Tokyo, Cir­ca 1910, Get Brought to Life with Arti­fi­cial Intel­li­gence

Watch Scenes from Belle Époque Paris Vivid­ly Restored with Arti­fi­cial Intel­li­gence (Cir­ca 1890)

A Trip Through New York City in 1911: Vin­tage Video of NYC Gets Col­orized & Revived with Arti­fi­cial Intel­li­gence

Icon­ic Film from 1896 Restored with Arti­fi­cial Intel­li­gence: Watch an AI-Upscaled Ver­sion of the Lumière Broth­ers’ The Arrival of a Train at La Cio­tat Sta­tion

Ayun Hal­l­i­day is an author, illus­tra­tor, the­ater mak­er and Chief Pri­ma­tol­o­gist of the East Vil­lage Inky zine. Fol­low her @AyunHalliday.

Watch Vintage Footage of Tokyo, Circa 1910, Get Brought to Life with Artificial Intelligence

For more than 200 years, the rulers of Japan kept the coun­try all but closed to the out­side world. In 1854, the “Black Ships” of Amer­i­can com­man­der Matthew Per­ry arrived to demand an end to Japan­ese iso­la­tion — and a com­mence­ment of Japan­ese world trade. With­in decades, many fash­ion-for­ward Euro­peans and even Amer­i­cans could­n’t get enough things Japan­ese, espe­cial­ly the art, crafts, and cloth­ing that exem­pli­fied kinds of beau­ty they’d nev­er known before. (Vin­cent van Gogh was a par­tic­u­lar­ly avid fan.) But if Japan changed the West, the West trans­formed Japan, a process ful­ly in effect in the footage above, shot on the streets of Tokyo between 1913 and 1915.

These scenes may look famil­iar to ded­i­cat­ed Open Cul­ture read­ers, and indeed, we pre­vi­ous­ly fea­tured anoth­er ver­sion of this film back in 2018. With its speed cor­rect­ed to remove the herky-jerk­i­ness com­mon to old films and with back­ground noise added, these glimpses of the men, women, and many chil­dren of the Japan­ese cap­i­tal, all of them liv­ing between the inward-look­ing tra­di­tion of their coun­try as it had been and the onrush of moder­ni­ty from with­out, already felt real­is­tic.

But now you may feel you’ve been per­son­al­ly trans­port­ed to this cul­tur­al­ly and eco­nom­i­cal­ly heady time in the Land of the Ris­ing Sun thanks to the work of Denis Shiryaev, a Youtu­ber who spe­cial­izes in enlarg­ing and restor­ing vin­tage film clips with arti­fi­cial intel­li­gence.

Shi­rayev is also respon­si­ble for the enhanced ver­sions of scenes from Belle Époque Paris, czarist Moscow, Vic­to­ri­an Eng­land, New York City in 1911, and even the Lumière Broth­ers’ ear­ly motion pic­ture The Arrival of a Train at La Cio­tat Sta­tion. At the begin­ning of this video he reveals the stages of the process that brought this cen­tu­ry-old footage of Tokyo to greater vivid­ness: de-nois­ing and dam­age removal, col­oriza­tion, facial restora­tion, and upscal­ing to 4K res­o­lu­tion at 60 frames per sec­ond — all assist­ed by neur­al net­works that, “trained” on rel­e­vant visu­al mate­ri­als new and old, crisp and weath­ered, to deter­mine the best ways to make it all look more con­vinc­ing. The results may make you won­der what else will soon be pos­si­ble — sure­ly not a feel­ing unknown to  these ear­ly 20th-cen­tu­ry Toky­oites.

Relat­ed Con­tent:

Watch Life on the Streets of Tokyo in Footage Record­ed in 1913: Caught Between the Tra­di­tion­al and the Mod­ern

A Trip Through New York City in 1911: Vin­tage Video of NYC Gets Col­orized & Revived with Arti­fi­cial Intel­li­gence

Time Trav­el Back to Tokyo After World War II, and See the City in Remark­ably High-Qual­i­ty 1940s Video

Down­load Hun­dreds of 19th-Cen­tu­ry Japan­ese Wood­block Prints by Mas­ters of the Tra­di­tion

1850s Japan Comes to Life in 3D, Col­or Pho­tos: See the Stereo­scop­ic Pho­tog­ra­phy of T. Ena­mi

The Entire His­to­ry of Japan in 9 Quirky Min­utes

Based in Seoul, Col­in Mar­shall writes and broad­casts on cities, lan­guage, and cul­ture. His projects include the book The State­less City: a Walk through 21st-Cen­tu­ry Los Ange­les and the video series The City in Cin­e­ma. Fol­low him on Twit­ter at @colinmarshall, on Face­book, or on Insta­gram.

16th Century Bookwheels, the E‑Readers of the Renaissance, Get Brought to Life by 21st Century Designers

Most of us, through our com­put­ers or our even our phones, have access to more books than we could ever read in one life­time. That cer­tain­ly would­n’t have been the case in, say, the mid­dle ages, when books — assum­ing you belonged to the elite who could read them in the first place — were rare and pre­cious objects. Both books and lit­er­a­cy became more com­mon dur­ing the Renais­sance, though acquain­tance with both could still be con­sid­ered the sign of a poten­tial­ly seri­ous schol­ar. And for the most seri­ous Renais­sance schol­ars of all, Ital­ian mil­i­tary engi­neer Agosti­no Ramel­li designed the book­wheel, an elab­o­rate mechan­i­cal device allow­ing the user to turn from one book to anoth­er in rel­a­tive­ly quick suc­ces­sion.

First drawn by Ramel­li in 1588 (and pre­vi­ous­ly fea­tured here on Open Cul­ture in 2017) but nev­er actu­al­ly con­struct­ed by him, the book­wheel has attract­ed renewed atten­tion in the 21st cen­tu­ry. “In 2018, a group of under­grad­u­ate engi­neer­ing stu­dents at the Rochester Insti­tute of Tech­nol­o­gy set out to build two,” writes Atlas Obscu­ra’s Claire Voon. “They began by dili­gent­ly study­ing the Ital­ian engineer’s illus­tra­tion, then pro­cured his­tor­i­cal­ly accu­rate mate­ri­als, such as Euro­pean beech and white oak.

With the help of mod­ern pow­er tools and process­es, such as com­put­er mod­el­ing and CNC rout­ing, they brought it to life.” You can see the RIT book­wheels under con­struc­tion and in action in the video above. (Its schemat­ics, near-impos­si­bly com­plex by the stan­dards of Ramel­li’s day, are also avail­able at RIT’s web site.)

Oth­ers have also brought Ramel­li’s design into real­i­ty. In the video just above, for exam­ple, we have writer Joshua Foer (pre­vi­ous­ly fea­tured here for his work on the sci­ence of mem­o­riza­tion) tak­ing his own repro­duc­tion for a spin. “It’s a fer­ris wheel for books,” Foer explains, “so that a schol­ar can have eight books in front of them, sort of like tabbed brows­ing before tabbed brows­ing.” The device’s cher­ry wood and laser-cut gears are cer­tain­ly hand­some, but what of its prac­ti­cal­i­ty? “I often read mul­ti­ple books at one time, and this way I can have them all open in front of me.” Most all of us start more books than we can fin­ish, and as we attempt to read them all in par­al­lel, occa­sion­al­ly one or two do get for­got­ten. Hence one advan­tage, even in our mod­ern times, of Ramel­li’s book wheel: any book placed on it becomes as unig­nor­able as the machine itself.

Relat­ed Con­tent:

Behold the “Book Wheel”: The Renais­sance Inven­tion Cre­at­ed to Make Books Portable & Help Schol­ars Study (1588)

Dis­cov­er the Jacobean Trav­el­ing Library: The 17th Cen­tu­ry Pre­cur­sor to the Kin­dle

The Art of Mak­ing Old-Fash­ioned, Hand-Print­ed Books

Won­der­ful­ly Weird & Inge­nious Medieval Books

Wear­able Books: In Medieval Times, They Took Old Man­u­scripts & Turned Them into Clothes

How to Mem­o­rize an Entire Chap­ter from “Moby Dick”: The Art and Sci­ence of Remem­ber­ing Every­thing

Based in Seoul, Col­in Mar­shall writes and broad­casts on cities, lan­guage, and cul­ture. His projects include the book The State­less City: a Walk through 21st-Cen­tu­ry Los Ange­les and the video series The City in Cin­e­ma. Fol­low him on Twit­ter at @colinmarshall, on Face­book, or on Insta­gram.

The End of an Era: A Short Film About The Last Day of Hot Metal Typesetting at The New York Times (1978)

This is usu­al­ly what hap­pens when I write a piece for Open Cul­ture: As I drink an over­priced cof­fee at my local cof­fee shop, I research a top­ic on the inter­net, write and edit an arti­cle on Microsoft Word and then copy and paste the whole thing into Word­Press. My edi­tor in Open Cul­ture’s gleam­ing inter­na­tion­al head­quar­ters up in San Fran­cis­co gives it a look-over and then, with the push of a but­ton, pub­lish­es the arti­cle on the site.

It’s sober­ing to think what I casu­al­ly do over the course of a morn­ing would require the effort of dozens of peo­ple 40+ years ago.

Until the 1970s, with the rise in pop­u­lar­i­ty of com­put­er type­set­ting, news­pa­pers were print­ed the same way for near­ly a cen­tu­ry. Lino­type machines would cast one line at a time from molten lead. Though an improve­ment from hand­set type, where print­ers would assem­ble lines of type one char­ac­ter at a time, lino­type still required numer­ous skilled print­ers to assem­ble each and every news­pa­per edi­tion.

The New York Times tran­si­tioned from that ven­er­at­ed pro­duc­tion method to com­put­er type­set­ting on Sun­day, July 2, 1978. David Loeb Weiss, a proof­read­er at the Times, doc­u­ment­ed this final day in the doc­u­men­tary Farewell — Etaoin Shrd­lu.

The title of the movie, by the way, comes from the first two lines of a printer’s key­board, which are arranged accord­ing to a letter’s fre­quen­cy of use. When a print­er typed “etaoin shrd­lu,” it meant that the line had a mis­take in it and should be dis­card­ed.

Watch­ing the movie, you get a sense of just how much work went into each page and how print­ers were skilled crafts­men. (You try spot­ting a typo on a page of upside down and back­wards type.) The film also cap­tures the furi­ous ener­gy and the cacoph­o­ny of clinks and clanks of the com­pos­ing room. You can see just how much phys­i­cal work was involved. After all, each page was print­ed off of a 40-pound plate made of lead.

The tone of the movie is under­stand­ably melan­choly. The work­ers are bid­ding farewell to a job that had exist­ed for decades. “All the knowl­edge I’ve acquired over my 26 years is all locked up in a lit­tle box now called a com­put­er,” notes one print­er. “And I think most jobs are going to end up the same way.” Some­one else wrote the fol­low­ing on the com­pos­ing room’s chalk­board. “The end of an era. Good while it last­ed. Cry­ing won’t help.”

You can watch the full doc­u­men­tary above. It will also be added to our list of 200 Free Doc­u­men­taries, a sub­set of our meta col­lec­tion, 4,000+ Free Movies Online: Great Clas­sics, Indies, Noir, West­erns, Doc­u­men­taries & More.

Note: This post orig­i­nal­ly appeared on our site in August 2015.

Relat­ed Con­tent:

Dis­cov­er the Inge­nious Type­writer That Prints Musi­cal Nota­tion: The Keaton Music Type­writer Patent­ed in 1936

Friedrich Nietzsche’s Curi­ous Type­writer, the “Malling-Hansen Writ­ing Ball”

The Art of Col­lo­type: See a Near Extinct Print­ing Tech­nique, as Lov­ing­ly Prac­ticed by a Japan­ese Mas­ter Crafts­man

The Endur­ing Ana­log Under­world of Gramer­cy Type­writer

Mark Twain Wrote the First Book Ever Writ­ten With a Type­writer

Jonathan Crow is a Los Ange­les-based writer and film­mak­er whose work has appeared in Yahoo!, The Hol­ly­wood Reporter, and oth­er pub­li­ca­tions. You can fol­low him at @jonccrow. And check out his blog Veep­to­pus, fea­tur­ing lots of pic­tures of vice pres­i­dents with octo­pus­es on their heads.  The Veep­to­pus store is here.

Artificial Intelligence Brings to Life Figures from 7 Famous Paintings: The Mona Lisa, Birth of Venus & More

Denis Shiryaev is an AI wiz­ard who has lib­er­al­ly applied his mag­ic to old film—upscaling, col­oriz­ing, and oth­er­wise mod­ern­iz­ing scenes from Vic­to­ri­an Eng­land, late Tsarist Rus­sia, and Belle Époque Paris. He trained machines to restore the ear­li­est known motion pic­ture, 1888’s Round­hay Gar­den Scene and one of the most mythol­o­gized works of ear­ly cin­e­ma, the Lumière Broth­ers 50-sec­ond Arrival of a Train at La Cio­tat Sta­tion.

Shiryaev’s casu­al dis­tri­b­u­tion of these efforts on YouTube can make us take for grant­ed just how extra­or­di­nary they are. Such recre­ations would have been impos­si­ble just a decade or so ago. But we should not see these as his­toric restora­tions. The soft­ware Shiryaev uses fills in gaps between the frames, allow­ing him to upscale the frame rate and make more natur­is­tic-look­ing images. This often comes at a cost. As Ted Mills wrote in an ear­li­er Open Cul­ture post on Shiryaev’s meth­ods, “there are a lot of arti­facts, squooshy, mor­ph­ing moments where the neur­al net­work can’t fig­ure things out.”

But it’s an evolv­ing tech­nol­o­gy. Unlike wiz­ards of old, Shiryaev hap­pi­ly reveals his trade secrets so enter­pris­ing coders can give it a try them­selves, if they’ve got the bud­get. In his lat­est video, above, he plugs the NVIDIA Quadro RTX 6000, a $4,000 graph­ics card (and does some grip­ing about rights issues), before get­ting to the fun stuff. Rather than make old film look new, he’s “applied a bunch of dif­fer­ent neur­al net­works in an attempt to gen­er­ate real­is­tic faces of peo­ple from famous paint­ings.”

These are, Shiryaev empha­sizes, “esti­ma­tions,” not his­tor­i­cal recre­ations of the faces behind Leonardo’s Mona Lisa and Lady with an Ermine, Botticelli’s mod­el for The Birth of Venus, Vermeer’s for Girl with a Pearl Ear­ring, or Rembrandt’s The Night Watch. In the case of Amer­i­can Goth­ic, we have a pho­to of the mod­el, artist Grant Wood’s sis­ter, to com­pare to the AI’s ver­sion. Fri­da Kahlo’s Self-Por­trait with Thorn Neck­lace and Hum­ming­bird gets the treat­ment. She left per­haps a few hun­dred pho­tographs and some films that prob­a­bly look more like her than the AI ver­sion.

The GIF-like “trans­for­ma­tions,” as they might be called, may remind us of a less fun use of such tech­nol­o­gy: AI’s abil­i­ty to cre­ate real­is­tic faces of peo­ple who don’t exist for devi­ous pur­pos­es and to make “deep fake” videos of those who do. But that needn’t take away from the fact that it’s pret­ty cool to see Botticelli’s Venus, or a sim­u­la­tion of her any­way, smile and blink at us from a dis­tance of over 500 years.

Relat­ed Con­tent:

Watch Scenes from Czarist Moscow Vivid­ly Restored with Arti­fi­cial Intel­li­gence (May 1896)

Watch AI-Restored Film of Labor­ers Going Through Life in Vic­to­ri­an Eng­land (1901)

Icon­ic Film from 1896 Restored with Arti­fi­cial Intel­li­gence: Watch an AI-Upscaled Ver­sion of the Lumière Broth­ers’ The Arrival of a Train at La Cio­tat Sta­tion

The Ear­li­est Known Motion Pic­ture, 1888’s Round­hay Gar­den Scene, Restored with Arti­fi­cial Intel­li­gence

Josh Jones is a writer and musi­cian based in Durham, NC. Fol­low him at @jdmagness

« Go BackMore in this category... »
Quantcast