A while ago, I attended the farewell/retirement party for my friend Ed Covannon. You may remember Ed's posts in the blog-formerly-known-as-1000Nerds about China & Innovation. Held at a social establishment across the street from our previous work location (reduced to a pile of rubble, now cleaned up), the roast was emcee'd by another former co-worker who chooses to remain nameless.  Let's call him "Mr. X" - and since we're all friends here, we'll just call him "X". Ed's farewell was a great multi-media occasion. This, of course, is what happens when a bunch of nerds gather together to pay homage to one of our own. (We'll spare you the haiku tributes, published internally under the moniker "Edless".)

Buildings 65 & 69 being imploded. 
The referenced social establishment is to the right, just out of the picture
Photo credit to John Dyer.

So during "X's" portion of the roast, he divulged a bit of information that was heretofore unknown to most people. While I initially thought he was kidding, I soon realized that a fraud had been perpetrated on the 1000Nerds blog! The picture of Ed that we had so faithfully published with his posts was, in fact, FAKE - it was a synthesized image! Furthermore, "X" was the mastermind behind the creation!

Who knew? When I first received Ed's picture to set up his blog profile, I thought the picture was quite striking. And I did comment to him that he bore quite a resemblance to the guy from Schindler's List... you know, Liam Neeson. Well Ed thought that the jig was up at that point - that he had been found out - but that wasn't the case at all. Ed is a handsome man in his own right. And while it's true that (being the artistic, scientific, creative Lithuanian that he is) he went through a bit of a wild phase, you'd never know that now. I felt that the picture was a nice, artistic interpretation of the inner Ed - smoldering, introspective, ... a really great picture for his profile. For reference, the before/wild and after pictures below were part of "X's" presentation.

Ed does have that classic look, one that is universal enough that he is often compared to any number of Hollywood leading men - most notably Harrison Ford and Liam Neeson. The peak of the Ed-look-alike phenomenon took place when the James Bond movie starring Daniel Craig hit the screens. What was fascinating was the number of people who thought that the new Bond looked just like Ed (not even the other way around).

Now look-alikes are not new. Wikipedia has a brief article on the matter. In fact, a quick Google search finds over 3 million websites covering look-alikes. I would have never thought that Sean Connery and the Ayatollah Khomeini resemble each other. But, if you don't believe me, check this out. (We should note here that people tend to tell "X" that his look-alike is Danny DeVito.)

So the looming question, at least in the mind of our intrepid "X," was this: Could a virtual Ed Covannon with look-alike provenance be synthesized through the magic of digital imaging? Further, could this virtual Ed be synthesized ONLY using image content from his alleged look-alikes? The question is not: Why do it? That's easy - because you can! The important question then becomes:  How to do it.

Step 1: Finding a software package
Pictures can be processed using a simple morphing program to create linear combinations based on the input images, using them as orthogonal eigenfunctions (this part is for the Nerds). The program used for this exercise was Morph Studio. It's been discontinued, but a Google search on "morphing software" will show lots more candidates.

Step 2: Acquiring the input images
"X" googled Ed's look-alikes and searched for images of Ford, Neeson, and Craig having approximately the same pose. The images were then cropped and rendered in black white. These images would be processed with our simple morphing program. One certainly could see some of Ed in each of these pictures. But could the right combination of digital imagery DNA lead to Ed's likeness?

Step 3: Dialing in some representative combinations
Now for the fun part. The first linear combination was the 50-50 mix of Ford and Neeson shown below.

Perhaps there was a tad too much Ford in that recipe. Throwing in a pinch of Craig to kick it up a notch is shown below.

There was definitely some degree of convergence here.Too bad that there was less Craig than the others, what with the thrill of being James Bond and all. But a Han Solo and Qui-Gon Jinn with a pinch of Bond hybrid can't be all bad.

Step 4: Making the analytical choice
Several linear combinations were then synthesized and evaluated. The thought of non-linear combinations (i.e. including distortion) came to mind, but let's walk before we can run. This was quickly becoming too complex of an algorithm and frankly could consume a lot of time. So to cut to the chase, "X" recruited the consulting services of the ultimate authority - Ed's wife - figuring that, of all people, she would expedite convergence of this blossoming process. Donna's final selection is shown below. It wound up being a 55-30-15 mix of Neeson-Ford-Craig.

Voila! Lest you doubt the technique, here's a comparison to the real Ed.

So everyone was amused and "X's" endeavor in Digital Genetic Imagineering was fulfilled. Which brings us back to the beginning of the story. Unbeknownst to "X", Ed submitted the picture to the blog. Frankly, if you're going to synthesize your own picture, what better place to do it than on the Kodak Tech blog?And though I did comment on the picture, why would I doubt a standup guy like Ed?Actually, Ed had sworn "X" to secrecy since he was apparently intimidated by the potential "Wrath of Jane."

[This "Wrath of Jane" stuff makes me think this might be a good time to digress for a moment and make a statement once and for all... Engineering degree notwithstanding and in spite of what many of my co-workers have insinuated, I am NOT the model for the Alice character in Dilbert!  I don't even know Scott Adams! And the next guy who says it may well enjoy my FIST OF DEATH!]

But getting back to the story... A few blog readers have made comments to Ed about the picture: "that must have been a few years ago", or "did anyone tell you that you look like a young Harrison Ford?" The overwhelming reaction however, was acceptance - people not caring, not being observant enough, or a subliminal validation of our Virtual Ed. We humans tend to trust the obvious. Perhaps the virtual image was closer to our mental image of Ed than the actual picture! By now Virtual Ed has become Ed's trademark, his brand. ; And now you know whole story...

Ah, the magic of digital imaging. So give it a try folks. Perhaps the important question is not "What's in your wallet?" but "Who's in your face?"

 

Note: Kudos to Ed Covannon and "Mr. X" for their contributions to this post.