Visual Turing test

[Markbnj]

http://turing.deepart.io/

I got 6/10. Was harder than I expected.

 

[balloonshark]

I also got 6/10.

 

[Schmide]

I got 6/10
Need more resolution.

 

[Paratus]

7/10


I give it 5/7

 

[Imp]

7/10... Non-human ones used too much geometry and straight edges, it appears.

 

[Subyman]

6/10. I don't know if I would call this AI, but it seems like they just passed a photo through a filtering process.

 

[Lean L]

10/10. Just think of what images you can find on a google image search. Most of the PC generated ones lack something that I can't quite put my finger on.

Human drawn images lacked a certain realism since it isn't generated from a photograph. (Image set 7)

 

[Subyman]

10/10. Just think of what images you can find on a google image search. Most of the PC generated ones lack something that I can't quite put my finger on.

Human drawn images lacked a certain realism since it isn't generated from a photograph. (Image set 7)

Maybe you are a robot?

 

[HamburgerBoy]

7/10.

 

[UnklSnappy]

8/10
Got the 2nd and last one wrong.

 

[JEDIYoda]

10/10. Just think of what images you can find on a google image search.

so you cheated???

 

[Lean L]

so you cheated???

Not what I was saying at all.

All the computer generated art starts with a base image, most likely something that you would see on google image search. Then it applies a predefined style to mimic art. That means if it looks like the base image is something you can find on an image search, that is the clue.

Honestly image set 6 and 7 took some observation but the rest were easy enough. Christ look at the sail boat on set 9... No artist would start a canvas with a sailboat then paint streaks through the background without overlapping the boat for style and the fact that it looks terrible.

 

[Matthiasa]

Got 7/10.

 

[Jeff7]

7/10.

#6 was a guess: One of the images wouldn't load.


Full disclosure: All were damn close to being guesses.
Back during primary school, my thought of painting was something like "There are cameras now. Why do we still have to paint things?"
My artistic ability hasn't changed much since then.:$

 

[SKORPI0]

7/10.

 

[Rakehellion]

Not what I was saying at all.

All the computer generated art starts with a base image, most likely something that you would see on google image search. Then it applies a predefined style to mimic art. That means if it looks like the base image is something you can find on an image search, that is the clue.

Honestly image set 6 and 7 took some observation but the rest were easy enough. Christ look at the sail boat on set 9... No artist would start a canvas with a sailboat then paint streaks through the background without overlapping the boat for style and the fact that it looks terrible.

The computer generated images look too homogenous. Human artists tend to highlight areas of the image and leave out details in others.

 

[Blitzvogel]

8/10

It comes down to patterning and how homogeneous the pattern is stylistically.

 

[mikeford]

Your result is 9/10!

 

[Jeff7]

The computer generated images look too homogenous. Human artists tend to highlight areas of the image and leave out details in others.

8/10

It comes down to patterning and how homogeneous the pattern is stylistically.

This itself is just another pattern. The next rev of the software might be able to address this.

 

[Red Squirrel]

5/10 but hard to tell at such low res, so it was mostly guess work lol.

 

[sandorski]

5/10 Mostly just guessed. The pics are so small it was difficult to really see any kind of pattern..... and I wasn't interested enough to take my time with it.

 

[Markbnj]

5/10 Mostly just guessed. The pics are so small it was difficult to really see any kind of pattern..... and I wasn't interested enough to take my time with it.

I suspect that was intentional, so that you would focus more on the style and an overall impression.

 

[John Connor]

5/10

 

[Rakehellion]

This itself is just another pattern. The next rev of the software might be able to address this.

But a computer doesn't know what part of the image a human would think is important.

 

[Jeff7]

But a computer doesn't know what part of the image a human would think is important.

A human looking at the painting might also not know what part of the image another human would think is important, especially someone doing something creative and artistic.

"Computer: Find a prominent shape or figure on this image. Use a 38% chance of employing a stylistic variation on that shape."

Something along those lines.