Context
The intended audience for this blog are people who are interested in how reality is portrayed. With daily cries of ‘Fake News’, neural networks swapping faces with people (see deep fake) and stylised depictions of people’s lives on social media, it is difficult for the untrained eye to tell the real and not-real apart.
The domain I have chosen is photography. When I’m not glued to a computer finishing assignments, I enjoy taking my camera out and about. In the world of digital photography, it’s almost standard practice to manipulate the photos you take to some extent. My own style is to make moderates adjustments to enhance and refine, but to keep it within the realms of what could be considered realistic. I am often fascinated by how excited people on social media especially, can get really excited over heavily edited images. I hope they know people aren’t always so good-looking, and the world is not so beautiful all the time.
Seeing is believing
This blog will demonstrate how the colour profiles of RAW image files can change dramatically from an original photo (as taken), to a final, public and published one. I have used principal components analysis on the pixels of the image, to visualise the change in three dimensional colour space (Red, Green and Blue channels), as two dimensions.
I’ve written more about this approach here.
One image is mine, and the other is courtesy of kai.boet (few people publicly share their before and after images).
Image 1 - Portrait by Kai
Originals - Before & After
Colour space visualisation - Before & After
Image 2 - Krissy & Michael by Me
Originals - Before & After
Colour space visualisation - Before & After
Insights
What is interesting to me when looking at the change in scatter plots, is how the colour variance widens, while becoming more even for both images. In the first image, colours are predominantly green and yellow, but after editing, the spectrum now includes a lot more blue, and a little bit more pink. For the second image, the reds in Krissy’s dress have been emphasised. There’s a lot more yellow after the colour temperature was warmed. The end result is a colour space that nicely covers dominant primary colours, but achieves a smooth transition between them as well.
Final thoughts
My conclusions from all of this are not your typical ‘insights’. For me, it has been an exercise in describing how the colour variance of photos change with editing, and in doing so, how reality has been transformed in something more pleasing to the eye. A little closer the ‘Data Art’ side of the spectrum, than your usual hard, actionable insights.