http://friendfeed.com/rooms/24-hour-application-challenge

"On Friday, October 24th, take a day off from school, from work, from the real world, and spend 24 hours writing an application from start to finish."

Are you going to participate Philipp? If so, what do you think you'll come up with?

Not sure yet but if so I would probably wanna pick something random, e.g. by flipping a dictionary :)

[cross-posted to Friendfeed :)]

Here's my creation: http://TurnYourNameIntoaFace.com

I just registered the domain some hours ago so if you can't see it yet, check out http://blogoscoped.com/temp/turnyournameintoaface/

Philipp

You according to you :
http://turnyournameintoaface.com/face/09050604.png

just type "Philipp Lenssen" in the site you created

and don't support girls names

But how did you all these faces? Generated?

My Picture looks like me...weeeeird...

Do you see a similarity between Obama and Biden?

http://turnyournameintoaface.com/?name=Obama
http://turnyournameintoaface.com/?name=Biden

> and don't support girls names
>
> But how did you all these faces? Generated?

Well, it has both male and female shapes, but they are not directly connected to a the gender of the name, because even though I have a name/ gender database it wouldn't scale to say Chinese names, and I wanted to include those.

There are at the moment 4 base types, main face, nose, eyes and mouth. Each of these comes in 18 variations which I think means there are 104,976 possible combinations. The string you enter is checked and from this text, a face is constructed. So every name has exactly one face representation (though one face cannot be reverse-read to get to the name). It's a bit of an experiment in automatic avatar generation, a face could be created based on IP... URL... and so on.

George R

http://turnyournameintoaface.com/?name=Joe+Biden
http://turnyournameintoaface.com/?name=Barack+Obama

Nope :P

Also, lets throw another person with media attention into this...Nice goatee!
http://turnyournameintoaface.com/?name=Joe+The+Plumber

Funny, http://turnyournameintoaface.com/?name=Bill+Gates looks pretty screwed up.

http://turnyournameintoaface.com/face/09100815.png
Bush

http://turnyournameintoaface.com/face/08150809.png
George W. Bush

Philipp's string-to-avatar converter is an example of an Identicon. Other examples are Wavatars:
http://www.shamusyoung.com/twentysidedtale/?p=1462
and Visiglyphs:
http://digitalconsumption.com/forum/Visiglyphs-for-IP-visualisation
and Don Park's original graphical identicons:
http://haacked.com/archive/2007/01/22/Identicons_as_Visual_Fingerprints.aspx

I wonder how long before we see automatically-generated icons on the forum pages in place of Philipp's masterful hand-drawn icons. Gotta automate, to leave time for the blogging :-)

Oh, and MonsterID (scroll down to the comments):
http://scott.sherrillmix.com/blog/blogger/wp_monsterid/

Great links Roger! Brainstorming... what are some ways to create gender-neutral avatars (that would probably be necessary for the forum use case)?
- plants
- certain misc. objects
- abstract shapes, just colors
- monsters, robots
- animals
- motion (e.g. an abstract shape moving a certain recognizable way, in a loop)
- ...?

And what are some use cases of gender-specific auto generated avatars?
- populating a game
- generating e.g. 10,000 random ones and offer one to be picked upon signup by user
- .....?

It's not just gender that some people will care about. Some people won't want an avatar with a certain skin color, or one that's too fat, too thin, too punky, etc.

Don Park's original identicons occasionally generated geometric shapes that looked a little too much like swastikas for some people's taste, although he seems to have modified the algorithm since then.

It's possible to give the user some choice while still using automatic avatar generation. One way is to generate eight different candidate avatars (e.g. by adding the digit 1-8 to the name before generating each avatar) and letting the user select whichever they like best from a drop-down box. Of course the chosen digit is one more thing that needs to be remembered in the database or in the user's browser cookie.

What some sites do is to generate a random identicon. Then, if the user doesn't like their automatic icon, they can go to the Gravatar site
http://en.gravatar.com/
and set up an avatar of their choice which will be used instead of the generated icon.

Each gravatar is associated with an email address, but in fact all that's passed to the gravatar site is an MD5 hash, so for this forum you could just append "example.com" to the poster's name before taking the MD5 hash.

[Edit: Corrected explanation to read 1, 5, 9, 13 instead of 1, 4, 8, 12, for the second instance of numbers as well.]

Philipp
The idea and the first results are really fine

http://bayimg.com/image/califaabk.jpg

ok guys, which one of you did it?

IBM was granted a US patent for generating the gender of avatars based on the gender of a user's name.

"System for using gender analysis of names to assign avatars in instant messaging applications"
http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=%2Fnetahtml%2FPTO%2Fsrchnum.htm&r=1&f=G&l=50&s1=7,447,996.PN.&OS=PN/7,447,996&RS=PN/7,447,996
  
slashdot coverage: http://yro.slashdot.org/article.pl?sid=08/11/04/2221208

IBM's patent is of no practical use, because the internet is global.

It can use an algorithm to deduce that, within the context of English-speaking Americans, Teri is female and Terry is male.

But it doesn't know that in other cultures it's common for males to be called Dominique, Jocelyn and Loryn.

Below are some quotes from the patent.

"There are thousands of languages in use today, which provides for a wide variety of given or first names."

"Generally, an English speaking person can correctly guess whether a particular individual whom he never met is female or male during a communications session based on the first name of that person, when it is provided. But when messaging with people having names originating from other cultures with which the person is not very familiar, for example, people with Indian or Chinese first names, the task of assessing gender may be more challenging."