These women do not exist. They each are a composite of about 30 faces
that I created to find out the current standard of good looks on the
Internet.
On the popular Hot or Not web site, people rate others’ attractiveness on a scale of 1 to 10.
An average score based on hundreds or even thousands of individual
ratings takes only a few days to emerge.
I collected some photos from the site, sorted them by rank and used SquirlzMorph to create multi-morph composites from them. Unlike projects like Face of Tomorrow or Beauty Check where the subjects are posed for the purpose, the portraits are
blurry because the source images are low resolution with differences
in posture, hair styles, glasses, etc, so that I could use only 36
control points for the morphs.
What did I conclude about good looks from these virtual faces? First,
morphs tend to be prettier than their sources because face asymmetries
and skin blemishes average out. However, the low score images show
that fat is not attractive. The high scores tend to have narrow faces.
I will leave it to you to find more differences and to do a similar
project for men.
Ruler Dogbert, dbabbitt, Andrew*, Easy Pickins, and 415 other people added this photo to their favorites.
View 20 more comments
Melponeme_k 52 months ago | reply
This is a subject that interests me a lot. I've recreated faces in virtual figures for the game second life.
Even the most beautiful women have a certain amount of asymmetry. It depends on the degree. I don't think our eyes trust true symmetrical faces. Which is why I think the higher rated faces look false and too fantasy driven. But the face of 7.0, has just enough balance to make it appear real and very attractive.
strenoth 48 months ago | reply
Alright, if some one wanted to make this scientific, going back to ground zero.
Filter all the pictures to be exclusively face pictures. If you did this originally, you failed to mention it. This filters out deviations like shown off body parts or clothing boosting ratings, or frumpy clothes dropping them.
Next, create multiple different averages at each level, using different, smaller sets. Create some variances, post them side by side to compare. Maybe create a few ethnic-specific averages, ie average a bunch of African women, a bunch of Asian women, etc.. Maybe create a specific balanced racial average (ie, an average image with the same number of caucasian, hispanic, african, and asian woman)
Now, once you have them organized for yourself, drop them all back into hot or not. Let them sit for at least 2 months. See what they rate as after that. there will most likely be a an adjustment up for all the faces rated low.
post everything side by side, ei, the first assortment of average faces, then the order they are after Hot or Not is done with them.
not perfectly accurate, but it's a data set. Hypothesis can be drawn from any valid data set. Then you have to test it, ie, make a prediction, see if you predict correctly.
What I described, and the smaller test done, are both data-gathering experiments, "what happens if I..." not hypothesis testing experiments, which are "If I do A to B with C, then result D should happen. Let's see if I'm right."
Bogota Colombia 47 months ago | reply
Wow !!!!
two
Curious Sarah 44 months ago | reply
I love this idea. Some of them look a bit like valerie bertinelli.
alxflickrrr 43 months ago | reply
Nice idea. Thanks for sharing!
caseorganic 42 months ago | reply
Thanks for this. I've also done some HotorNot experiments, but nothing like this. Well done.
schietti 41 months ago | reply
great initiative... it's a relevant project, no doubt about it.
tavshande 38 months ago | reply
Very cool, There is now an iPhone app that measures how attractive someones face is on a scale from 1-10. It works for guys and girls.
www.tavshande.com/vanity/
itunes.apple.com/us/app/vanity/id318044093?mt=8
there is also a website with scores from celebrities:
www.thevanityreport.com
Tiger Woods is the latest celeb to be rated by vanity.
kill4hum 38 months ago | reply
Not surprising, but still very interesting. Do one for guys.
Larry Moran 36 months ago | reply
Well done and interesting. : )
ALiebe 32 months ago | reply
Hello Pierre,
These are incredible. I was wondering if I could ask you for permission to use the most attractive one and the least attractive one for a college psychology project. I would credit you by name or by flickr name and website or whatever you would like, so that if the article was presented, you would receive credit.
Allie
lilysleeper 31 months ago | reply
Look at their eyes. They steadily get bigger as you go from less attractive to more attractive.
megadosyanet 30 months ago | reply
harika bir resim olmus gercekende süermis
World News
Metin2 Hilesi
Tek Link
79d0bac0434a5f6a180581d1c6a6351a 29 months ago | reply
Dear photographer.
I am a researcher in facial recognition and am doing a survey on attractivenss. Your pictures would surve my purpose perfectly. I Know I can freely use your pictures if I give credit to you. So my question is how do you want te get crediter (format) user name on flickr is a bit informal
crestina.martins 26 months ago | reply
interesting
Dragan* 23 months ago | reply
Seen in: Fine Gold Post 1 - Award 5
4 Plus Awards Post Here
6 Plus Awards Post Here
cbeauche 19 months ago | reply
Un article aujourd'hui avec ta photo en exemple:
www.wired.com/wiredscience/2011/11/equations-for-geeks/?p...
Veronica Kai 18 months ago | reply
I love this.
a.j.schouten 15 months ago | reply
Hea Pierre, question - I am part of a social neuroscience research group and we are doing
an experiment on social comparision and attractiveness. We also use face morphs to acces different degrees of attractiveness. We are now in the stadium of finding more raw (rated) data to use in our morphs. Do you still have acces to the faces you used in your morphs? If you do, are you willing to share them? Please contact us at a.j.schouten@students.uu.nl
Karlaofallpeople 5 months ago | reply
I love how racism, sexism and casual dismissal of personality are so well balanced here. Awesome job