Betaface, He's Got The Cutest Little Betaface

In and after 1966--when we didn't have the luxury of the internet with its thousands of photos of The Beatles--Beatles fans saw a distinct difference between our Paul and the curious-faced man The Beatles introduced as Paul in late 1966.

A staple of all discussions of whether our Paul was replaced is comparisons of photos of the man of late 1963 through mid-1966 that American fans knew as Paul, and this curious-faced man.

Face recognition software is augmenting our ability to see differences between individuals.

So, I went looking on the internet and found three face recognition websites that are easy to access and easy to use:  1.)  www.pictriev.com 
                           2.)  www.betaface.com
                           3.)  www.twinsornot.net

I uploaded photos of our Paul and the last Paul replacement (1966-on) and found the following:

Betaface #1

Betaface #2

Pictriev #1

Pictriev #2

TwinorNot #1

TwinsorNot #2

                        
 As you can see, betaface and pictriev were reporting that the two Pauls were NOT similarly looking and were NOT the same man.  [Note:  The second face comparison on Betaface #2 is between the real Paul McCartney and his last replacement and Betaface says they also are NOT the same man.)

TwinsorNot consistently said they were the same man.  What's going on here?

TwinsorNot is a project of Microsoft.  The inventor of the app bragged that he developed it in 15 minutes--which doesn't reflect on its accuracy, if it is accurate.  One problem might be that it analyzes a tight square of the face which--in Faul's case--leaves out that famous long face that pinched-nose John sang about in "I Am The Walrus."  Remember?:  "Man you've been a naughty boy, you've let you're face grow long." 

But I took twinsornot's comparisons a step further when I found a stellar little project that a French-Canadian photographer is developing.  The photographer's name is Francois Brunelle and his project is called "I'm Not a Look-Alike!"

He started in Quebec photographing pairs of persons who look strikingly similar but are totally unrelated.  I found an internet article at the Mirror newspaper site:  www.mirror.co.uk of November 18, 2013 describing Brunelle's work and providing a gallery of some of Brunelle's "twins" photographs.  I chose photos numbers 1, 6, 7, 8 and 14:  "twins" I thought looked most alike.  I uploaded the images to twinsornot for its opinion. 

And--well, golly--the program thought four of the five pairs were twins.  (See below.)

 Remember, these are UNrelated persons.

The one pair twinsornot gave a less than 100% assessment on was photograph #6 of young girls:  they were given an 82% likeness.  (See below).

Compare twinsornot's photo of Paul and of Faul with its 100% likeness assessment and the photo of the young girls with the 82% likeness assessment and you can understand that some software has a better handle on matching likenesses [in this case, pictriev and betaface]----and so do many current-Paul doubters, included among them that gang of boppers who were seeing a new Paul in 1966.