Say goodbye to criticism, technology is here to rescue us!

[le vrai rdu]

2,9 /100 [:rofl]

 

[dazedgonebye]

My "best" shots get sub 20 scores. My most ordinary are all over 60.

It seems to like simple things with pretty colors.

 

[uhligfd]

How different are the judgement criteria at competitions? I generally do not score well with what I think is my best pic; just the second or third best will do ok.

Though, when for sale, the pics I like best get sold most often ...

Maybe that is how competitions have been judged for a long time ...

And the judges will/can now be replaced by computer ... as maps and directions have been by GPS. Brave New World!

 

[David William White]

Woohoo, I got a 60 with this one:

 

[ampguy]

This one got 69

This one got 69

from a half dozen I submitted from yesterday's photos taken:

 

[ampguy]

That's a good photo

That's a good photo

white balance is good, and you should get extra points for not cropping out the top and bottom edges. The duck probably added 20+ points. Very well done!

this is hysterical. This picture of mine got a score of 82.3. It's a shot I took years ago when preparing for a moving sale. The vase with fake flowers was listed at $4. Maybe I can print it out at 16x24 put it up in a gallery and call it Vase on Back of Toilet and ask $1,250 for it.

 

[Larky]

What a load of crap. The system I mean, not the picture. 🙂

 

[imajypsee]

I think you're conflating the notion

I think you're conflating the notion

that this thing is doing some sort of aesthetic or critical assessment of a photograph. It's not. It's trying to see if it can agree with human assessment on a mass scale. Human assessment on a mass scale is otherwise known as "popularity," which is not necessarily the same thing as an educated critique. The Brooklyn Museum recently did a "crowd sourced" online assessment of some photos made by regular folk which would then be exhibited in a gallery at the museum. The photos were even displayed much like the "tags" you see on Flickr, size accorded to sameness of use. Guess, what? The photos all pretty much agreed on what constitutes a "good" photo of the subject, which was "the changing faces of Brooklyn." http://www.nytimes.com/2008/07/04/arts/design/04clic.html

This computer rating thing is, indeed, dumb. But, at least I get to see some decent photos in this thread 😛

Mary in SW Florida, USA

 

[rover]

67.7

78.6

And I simply don't agree with this one. I am much better looking than a 38.7.

 

[Dave Wilkinson]

67.7

78.6

And I simply don't agree with this one. I am much better looking than a 38.7.

I'm pleased that first portrait of Roger scored well though!

 

[EliasK]

2.6/100 and 34.4/100


Poor Henri... 😀

 

[mich8261]

It seems to like simple things with pretty colors.

Steve, you knocking my flowers and rubber ducky 😉

 

[climbing_vine]

I was just about to say what Mary said, glad I finished reading the thread before replying to all the people who apparently didn't get what the point of the experiment is from the very simple one-sentence introduction. ;P

Personally, I think this is the coolest thing I've seen in weeks. I'm continually astonished and challenged by the mounting evidence that most of what humans think makes us so damn special (what picture is the prettiest? It's all about "human interaction!") can be emulated by simple prescriptions in a computer. It's pushing us to a deeper analysis of what, if anything, really does make us more special than the ants we kill in the kitchen.

I'm absolutely dumbfounded by people who don't think there's something enlightening about a computer program accurately predicting mass appeal.

 

[climbing_vine]

And, for the record, I'm utterly unaffected by every one of the "classic" pictures that have been posted here with low ratings in an attempt to prove that the machine is dumb. They're dull, muddled, and pointless. A french kiss? Big deal, I see them on the street every day. Watching thousands of them happen in different permutations and seeing what people are on about is infinitely more interesting than one poorly framed, crowd-interfering picture of one kiss one time. Boring!

 

[peterm1]

Hmmmmm. I presume the software takes account of various human factors - if so there is an old adage which is relevant. "No one ever went broke by underestimating the taste of the public." Say n' more

 

[climbing_vine]

Even computers hate parking meters.

 

[ferider]

I think it's pretty cool. Might change the cinematography of American Idol, for instance 🙂

Or the next election campaign.

In comparison, who cares about classic art ?

Also somehow funny, that while we give the expert system pictures to play, we are training it ...

Cheers,

Roland.

 

[Leighgion]

The algorithm appears to love my bridge photos.

91.9

And likes the same bridge shot on film slightly better than digital, despite vignetting.

96.9

On the other hand it doesn't like classic English cars shot on Velvia.

13

It hates American blues drummers in Europe shot on digital.

6.8

I don't think I buy even the modest claim that this system is doing much of a job predicting mass appeal. We are part of the masses too, and should be able to see some kind of visual argument for these ratings, but I'm just not getting it.

 

[dof]

This reminds me of a project done at MIT a few years ago called, EigenRadio. As I recall, the researchers there sampled the most most heard music on the radio within a set geographic area and combined the resulting sound into one meta-stream of audio. The result was statistically the most popular form of music one could encounter. Sonically it was a whole lot of noise. Fascinating stuff really.

Not sure if the link is down only temporarily or permanently, but Iwas unable to log on earlier today.

eigenradio.media.mit.edu

 

[Keith]

Someone show it some Noctilux photos at f1 and see if it appreciates premium bokeh and vignetting! 😛