What Did George Washington Look Like? A Computer Algorithm Will Show You

Posted on by Benjamin Chabot-Hanowell (Brash Equilibrium)
URL for sharing: http://thisorth.at/x86
36063
ScienceNOW, the utterly sensationalized offshoot of the otherwise reputable Science magazine, reported on a new computer algorithm that may help historians get a more accurate picture of past U.S. presidents. Literally. When you think of George Washington, you probably think of the craggy, solemn face you see on the one dollar bill (unless you are under the age of twenty, in which case you don't know who George Washington is, and you've never used paper currency; flame away, youngsters). The iconic portrait was painted by Gilbert Stuart, who also immortalized five other U.S. presidents. Physician and history buff Eric Altschuler, seeing these portraits in the Smithsonian's National Portrait Gallery in Washington, D.C., asked himself what we've all been thinking: Did those dudes really look like that?

Who is this?
Pop quiz: Who is this?
A. George Washington
B. That old dude from Up
C. All of the above
D. ur jst jely, uld man

Altschuler got some funding from the National Science Foundation and hired MIT genius graduate student Krista Ehinger to help him figure things out. The two devised a computer algorithm to do the grunt work. The algorithm, based on available photographs of three of the presidents, is supposed to strip Stuart's artistic signature from the dead executives faces, providing a more realistic image. The creators of the algorithm argue it can shed light on the visage of other presidents painted by Stuart, whose photographs are unavailable. Here's how it worked:

  1. Ehinger and Altschuer drew computer outlines of the matching photos and portraits.
  2. They measured the differences between the two computer outlines.
  3. They assumed those differences measured Stuart's departure from reality.
  4. They averaged all those differences between photo and portrait as a model for Stuart's artistic signature.
  5. Aided by their computer model, they altered the portraits of presidents whose photographs were unavailable.
  6. ???
  7. Profit.
I'm calling semi-bull$%&* on this for one reason. The photos the algorithm uses were taken long after the portraits were painted, when the presidents had become old and bitter. Hell, Barack Obama already looks seventy, and dude's been president only two years! Here's an example: Below are the photo, Stuart portrait, and algorithm-generated portrait of John Quincy Adams, our sixth president.

JQ Adams, boyyyyyy

All I gotta say is, "Come on!" No. I'll let Gob say it for me.



Below is an even better (or rather, worse) example, comparing the photo, Stuart portrait, and computer outlines of Daniel Webster. Webster wasn't a president, but he was a notable statesman (14th and 19th U.S. Secretary of State) during our great nation's Antebellum Period of westward expansion and Native American genocide.

Daniel Webster

Comparing Webster's photo and portrait as equivalent images of the man is like comparing young Marlon Brando to old Marlon Brando:


Marlon Brando: From sex god to Dr. WTF Moreau.

Nonetheless, we at ToT have, in the name of Science Most High, replicated this work on more recent presidential physiognomies (and the faces of other important historical figures). The results are notably freaky. No, we don't give a damn that we've completely bastardized the process, using images Stuart didn't paint. This is innovative, inter-disciplinary science. We don't have time to stop and thing about the nitty gritty details, okay?

Science is awesome.
This science is almost as awesome as the science we're gonna do.

First, we ran the algorithm on the face of Barack Obama, our current president. Oddly, the program seemed to mirror Americans' polarized view of Barry, thus illustrating the true genius of Ehinger and Altschuler's work.

Here's the image we used:

Barry.
Barry, before he realized the presidency would age him twenty years in a matter of two.

Here are the two pictures the algorithm spat out:

Barack is Jesus.

Barack is Hitler.
Apparently, Barack Obama is both Our Lord and Savior and Hitler, but nowhere in between.

Next, we ran George Bush through the algorithm. Here's the image we used:

George Bush
I'd have a beer with this guy.

Here's what the algorithm produced:

Trimmed Bush.
Shut up. We know you saw this coming.

Finally, we tried the algorithm on Bill Clinton. Here's the image we used:

God DAMN he's sexy!
Now that's a sexy man.

And the algorithm's result?

Clinton = White
No wonder!

Apparently, President Clinton was really Barry White all along. Suddenly, so many things make sense. We don't know why the algorithm inverted the image of our President and Walrus of Love. Perhaps that's how White/Clinton managed to fool us for so long. Honestly, I still would have voted for him, had I not been seven and then eleven years old when ran.

But does the algorithm work on non-political figures? To find out, we ran the algorithm on this image of Joan Rivers.

Joan Rivers before.

And this is what we got back:

Joan Rivers after.

Confused, we tweaked the algorithm a little bit, and ran the image again. This is what we got:

real Joan Rivers

The marvels of modern technology...


Does Altschuler and Ehinger's algorithm do what they say it does?

6063 views & 21 votes

Debate It! 3

Hysterical :)

Posted By The Other Thing,

I love scienceNOW! It helps me sound smart at parties. I'm not invited to a lot of parties anymore.

Posted By Actionamy,

I'm with Actionamy - love to watch ScienceNOW as well - while it does pander to the non-scientist, it has given me really good leads on interesting stuff for internet surfing - last week, among other things, I learned that Mac Doggie has a much better memory and can follow directions better than me and most monkeys :)

Posted By The Other Thing,

Make a Comment

You must be signed in to add a comment. login | register
Username
view profile
You are now following
a
You are no longer following
a
 
test message
×