I sanity-checked it with two headshots of the same third party and it swung ~2.5 points, so it seems to capture state (lighting/angle/expression) more than trait.
Then I uploaded my own photo and got an unexpectedly high score, which conclusively validates the model and my rigorous n=2 study.
hahaha yeah you mog
but for the first part; yeah it does take into account the angle, expression stuff, and like at first i was not ok with that but then i realised thats fine. cos like obv photos affect how attractive you look to a very high degree, like in some photos you may look like a 10/10 and in some you may look like at 5. so it should reflect that.
i think the main problem is, i only rate their face structures, cos thats the thing you can explicitly calculate; but obviously that's not all there is to how good you look.
also btw one of my friend, whos near the top read your comment and sent it to me with a crying emoji haha
I can't help but wonder if it's a bit of self-aware humour about "scientific looksmaxing". Symmetry is involved in attractiveness but symmetry alone can only carry you so far.
yeah i mean i kind of realised that you cant really rate faces based on a bunch of mathematical params and expect it to still be very accurate; but maybe my opinion of attractiveness is skewed
lol no one has hacked the website, those are the real ratings for those photos from my algo. people have found "bad" faces that get good ratings but maybe they're js not that bad haha
nono, i had only added like 10 images... i give people the option to add an instagram account but if they don't it defaults to the do_i_mog instagram account which is why most of the entries seem to come from do_i_mog
I made this site where you upload a photo of your face and it ranks your face on a bunch of measurable parameters , and then gives you a final rating based on how close your features are to the ideal face. There is also a public leaderboard.
This went kind of viral at my uni when i launched it, like over a 1000 people have gotten their ratings and people seem so exicted to see where there friends rank etc. curious to see what y'all think
The top-ranks on the leaderboard has me convinced this is a poor predictor of actual attractiveness to actual humans. You should have an option to privately rate a photo and not post it publicly. Seems like a big oversight.
The people high on that leaderboard are objectively ugly. Attractiveness is pretty consistent across individuals and cultures with only minor variations.
honestly the leaderboard was the most interesting part of the project for me. like people looking at their friends, the people near them who have similar rating etc. was the coolest part. if i allowed people to not post it publicly, most wont, and that would ruin the appeal of the site
i do think people sort of accepted the ratings cos
1) a lot of people with really high ratings were posting it on their instagram story
2) a lot of people tried to upload a lot of photos to try to get a higher rating
3) some people dm'd me sad messages like why do i have such a low rating haha
But, history has shown that people love to see how they rank up.
Historically, these sort of sites allowed users to rank each other.
Given yours is algorithm based, it'd make sense to give some context as to how the algorithm works... and an example of the 'ideal face' according to the algorithm.
sure, the algo basically is i calculate a bunch of objective details about your face (like angle for canthal tilt, distance ratios for mouth etc) , you can see these measurements in the overlaid images you get when you rate yourself. then according to the research online, i try to give a single score according to these measurements, for example, a positive canthal tilt is conventionally looked as more attractive. similarly less error in rule of fifths is seen as better or a specific face ratio as 0.42 is looked as the best, so i assign your mesasruemtns a specific single score according to this research. and then i combine these using some weights to get a final rating.
yeah feels like the facemash moment for my uni lmao
I sanity-checked it with two headshots of the same third party and it swung ~2.5 points, so it seems to capture state (lighting/angle/expression) more than trait. Then I uploaded my own photo and got an unexpectedly high score, which conclusively validates the model and my rigorous n=2 study.
hahaha yeah you mog but for the first part; yeah it does take into account the angle, expression stuff, and like at first i was not ok with that but then i realised thats fine. cos like obv photos affect how attractive you look to a very high degree, like in some photos you may look like a 10/10 and in some you may look like at 5. so it should reflect that.
The people at the top of the leaderboard aren't particularly attractive in my opinion.
i think the main problem is, i only rate their face structures, cos thats the thing you can explicitly calculate; but obviously that's not all there is to how good you look. also btw one of my friend, whos near the top read your comment and sent it to me with a crying emoji haha
I can't help but wonder if it's a bit of self-aware humour about "scientific looksmaxing". Symmetry is involved in attractiveness but symmetry alone can only carry you so far.
yeah i mean i kind of realised that you cant really rate faces based on a bunch of mathematical params and expect it to still be very accurate; but maybe my opinion of attractiveness is skewed
i think someone's already hacked the website
I made the same mistake when I vibecoded something for ShowHN and it was hacked within the hour lol
lol no one has hacked the website, those are the real ratings for those photos from my algo. people have found "bad" faces that get good ratings but maybe they're js not that bad haha
Those are added by the author themselves. Scrub the test data perhaps?
nono, i had only added like 10 images... i give people the option to add an instagram account but if they don't it defaults to the do_i_mog instagram account which is why most of the entries seem to come from do_i_mog
That's a nice way to put it.
bruh thats so rude lmaooo
I made this site where you upload a photo of your face and it ranks your face on a bunch of measurable parameters , and then gives you a final rating based on how close your features are to the ideal face. There is also a public leaderboard. This went kind of viral at my uni when i launched it, like over a 1000 people have gotten their ratings and people seem so exicted to see where there friends rank etc. curious to see what y'all think
The top-ranks on the leaderboard has me convinced this is a poor predictor of actual attractiveness to actual humans. You should have an option to privately rate a photo and not post it publicly. Seems like a big oversight.
lol, yikes, yeah at least start with a hot-or-not baseline as a sanity test.
people are subjective, math is objective ;)
The people high on that leaderboard are objectively ugly. Attractiveness is pretty consistent across individuals and cultures with only minor variations.
honestly the leaderboard was the most interesting part of the project for me. like people looking at their friends, the people near them who have similar rating etc. was the coolest part. if i allowed people to not post it publicly, most wont, and that would ruin the appeal of the site
This is fascinating... I am curious, are any concerns with the algorithm being raised? Or are people blindly accepting their ranking?
i do think people sort of accepted the ratings cos 1) a lot of people with really high ratings were posting it on their instagram story 2) a lot of people tried to upload a lot of photos to try to get a higher rating 3) some people dm'd me sad messages like why do i have such a low rating haha
hot or not is back.
I think the algorithm is janky based on the list.
But, history has shown that people love to see how they rank up.
Historically, these sort of sites allowed users to rank each other.
Given yours is algorithm based, it'd make sense to give some context as to how the algorithm works... and an example of the 'ideal face' according to the algorithm.
sure, the algo basically is i calculate a bunch of objective details about your face (like angle for canthal tilt, distance ratios for mouth etc) , you can see these measurements in the overlaid images you get when you rate yourself. then according to the research online, i try to give a single score according to these measurements, for example, a positive canthal tilt is conventionally looked as more attractive. similarly less error in rule of fifths is seen as better or a specific face ratio as 0.42 is looked as the best, so i assign your mesasruemtns a specific single score according to this research. and then i combine these using some weights to get a final rating. yeah feels like the facemash moment for my uni lmao