What were The Awesomeness Scores?

In my previous post on transforming photographs, I published four variations on this pretty photo of red tulips, musing that transformation should make them more attractive as Wall Art than this rather anonymous image.

this image and all on this page Copyright © 2017 D A Harcourt. All rights reserved 

Photographs like this are pretty good from a photographic point of view as the colour is nice, the bookha and depth of focus on the tulips is nice, there is a nice diagonal and there's a kind of a story in the height difference of the flowers - but isn't it like thousands of others online?

The images were processed using the Every Pixel Aesthetics Beta, an AI based program claimed to measure "the beauty of your picture" or it's "awesomeness", with the following results:

the original (A)

brightening it up (B)

dulling it down (C)

adding fantasy colors (D)

or turning up the heat (E)

That's rather interesting and seems to indicate a strong leaning towards brightness of the image. These few shots don't seem to discriminate between photograph like images and more graphic images which was something I'm interested in understanding.

One interesting check that happened by mistake was that I re ran the original image through the software almost 3 weeks later and found the score to be identical, to the decimal point! I will need to do more to confirm whether this is always the case, as it seems to me that you would expect small variations in a human like judgement process over such a period.

Email me with your thoughts!

Free Poster Files to Try and Use

DaveH Posterart is giving away free 1MP copies of the image files from which it prints its poster. Here is the first one which you will find on their website right now.

DaveH Photoart gives you the right to do whatever you wish with files so long as its not commercial, While the file is too small to produce large posters you could use a small print to see how it actually fits with your decor. The small file would be fine to produce items such as greeting cards, cellar phone covers and coffee mugs.

TRANSFORMATIONS TO IMPROVE POSTERS

In my focus on transforming my photographs into poster art I frequently ask the question "will this appeal" to poster buyers. (A)

this image and all on this page Copyright © 2017 D A Harcourt. All rights reserved. 

Photographs like this are pretty good from a photograph point of view as the colour is nice, the bookha and depth of focus on the tulips is nice, there is a nice diagonal and there's a kind of a story in the height difference of the flowers - but isn't it like thousands of others online?

So we set out to transform the photograph to make it more interesting, but the problem is there are so many potential ways of "improving" the photo - like brightening it up (B)

dulling it down (C)

 adding fantasy colors (D)

or turning up the heat (E)

But which one is going to be the most popular? Something I across the other day, may be a useful tool in this kind of choice.  EveryPixel is testing a "neural network" that judges the chance of an image "being awesome". I understand their raison d'être is to help users to weed out the useless images in stock photo sites.

This is how EveryPixel sees the first image in this post. Thats just to wet the appetite - I will be digging deeper next time but would appreciate your inputs, on anything about rating attractiveness, at anytime.

How would you rate the attractiveness of these 5 posters - email me using the reference letters.