Yesterday’s post featured this chart
It included one example photo for each quadrant. Today’s post shows all 68 photos.
Click on the above image for full screen (once doing that you may have to reduce it to fit your screen, “Ctrl -“ on a PC). Within each quad, the most popular image is at the top left and the least is at the bottom right.
Memorable, quads 2 & 3, vs. not as predicted by the MIT algorithm, is heavily biased by whether the image is a landscape or not.
- See this by comparing the number of landscapes on the left vs. right sides
What makes an image popular is less obvious. Especially since the MIT algorithm considered not only an image’s visual characteristics but, equally or more important, the social context in which it was viewed and by whom (in this case Flickr). Since I don’t shoot for a Flickr (WOW!!!!) audience, I’m not surprised that the MIT program and I don’t agree on a lot of its scores. That doesn’t say that either of us is right or wrong, just that we have different standards.
- Can anyone hazard an opinion as to what may separate images in the top half from those in the bottom.
This is what happens when you give an engineer a camera 😉