final word on the dress

Yesterday, I posted about The Dress that people see as either blue and black or white and gold. Following several radio and telephone interviews I wanted to have a final attempt to explain what is happening with the dress. It is quite an extraordinary phenomenon – yesterday the dress looked blue and black to me but my PhD student (looking at the same dress on the same screen) said it looked whitish and gold. When I came home last night and looked at the photo on my mac book, the same image that had looked decidedly blue and black to me before now looked whitish and gold. So what is happening?

The first thing is that it is nothing to do with the dress. The problem is with the photo of the dress. I believe that anyone looking at the dress in real life would certainly call it blue and black and also anyone looking at the manufacturer’s photo of the dress would also call it blue and black.

The second thing is that there is more than one phenomenon going on. The reasons why my PhD student and I saw different colours in my office may be a little different from the reasons why I saw it one colour on my pc in my office during the day and another colour during the evening on my mac. So, although people might like a simple answer and a soundbite, in my opinion the explanation is necessarily a little detailed. But I will try to avoid too much technical jargon below.

The camera does lie
I think many people believe that when they take a photograph and put in on the internet and people look at, what people are seeing is a faithful rendition of the original scene. People take this for granted, I believe, without giving it much thought. Unfortunately, this is not guaranteed. There are many reasons why the colour someone might look at in an image might not be the same one that was in the original scene. Different cameras capture colour in different ways depending upon the type of camera, the settings on the camera, and the light under which the image is taken, to name just three factors. In The Dress image, the image looks over-exposed and the colours are washed out. The black is quite pale and has a colour tint and the blue is very washed out and insipid. Hopefully you can see where this is going already.

Different displays show colour differently
You can put the same image on a PC, a mac, a smart phone and a tablet and look at it. The colours will probably not be identical. Reds will probably be red and blues will be blue. But the colours are likely to be not exactly the same on the different devices. If you are looking at your screen from an angle, the colours may change radically. Also, if you are looking at your screen in bright sunlight the colours may look more washed out – though some smart phones and tablets try to ‘intelligently’ correct for this which might make the problem better or worse. The fact that I saw the colours differently in my office than at home could be due to differences in the devices I was using or could be due to the lightening environment, The lighting in my office is quite different to that in my home, for example.

People see colour differently – a little bit
About 1 in 12 men are colour blind. Very few women are afflicted. But even for the rest of us – so-called normal observers – there is variability in our colour vision. One factor for this could be that there are known to be differences in our eyes from person to person. This effect could be small but may be a factor in this story. More important is probably the fact that if sit in a dark room for a while and get used to the dark our vision will be different to it would be if we were outside in bright sunshine. This so-called ‘adaption’ is one way our visual systems deal effectively with such a wide range of brightness from dark rooms to brightly illuminated outdoor scenes. Someone coming into a room from outside (where the sun and sky are very bright) might very well see different colours on the screen than some who had been in the room for a much longer period. These adaption factors are well known in science.

People don’t always agree on colour names
There are at least 3 million different colours in the world. How many colour names can you think of that we could broadly agree on? Words like, blue, black, red etc. There are others like beige and taupe where we might agree less well. But include these and how many do you have? 30? 50? 100? And these names have to cover 3 million colours!! So each name is a category that covers quite a large range of colours. Last year I published a paper where we gradually moved a colour from yellow to green and asked people to tell us when the colour went from yellow to green. Not surprisingly, the point at which people told us the name changed varied from person to person. So there are some colours that some people will call yellow and other people will call green. Correspondingly, just because two people are calling a colour by different names does not necessarily mean that they are seeing it as a different colour.

My final explanation
Variabilities in displays, viewing conditions, observers and colour-naming boundaries can cause disagreement in how to name colours. Normally, this would not shift a black to a gold or a blue to a white. However, in this case, the image that has caused the controversy is not a faithful reproduction of the original. Because of the way the image was taken the black has shifted considerably away from the centre of the category that we would call black. And likewise for the blue. In my office today I would still call it black. But it was not a strong convincing black. It was a little pale and had a bit of colour in it. To be honest, I could understand why someone else might call it gold. The colour was on the boundary between black and gold and now differences between people could cause it to be classified as one colour or the other. When I came home, the colours had shifted for me. I don’t think my colour naming boundaries had shifted. Rather, I think this was to do with the lighting I was viewing the colour in, or the screen (a mac rather than my pc) or the angle I was viewing my screen at. Any or all of these factors could have shifted the colour so that it passed from the category I call black to the one that I would call gold.

Maybe the surprising thing is that these controversies do not happen more. Colour imaging scientists have been phenomenally successful in delivering colour imaging devices that satisfy consumers. Part of this work is done at the University of Leeds where I work but there are other places around the world who make great contributions including RIT in Rochester USA. And then there are some super bright scientists in places like Samsung, Apple, HP and LG who have worked hard to understand the complexities of colour perception and colour communication to the extent that people barely even think about these issues. However, there is more work to be done. Colour is still a major factor in people being dissatisfied when they buy something over the internet. When the product arrives it is sometimes not the colour they expected it to be. And colour fidelity is still not good enough for many medical applications. If you want to get involved in colour science please contact me. My email is s.westland@leeds.ac.uk and you can also find me @stephenwestland

6 thoughts on “final word on the dress

  1. Hi Steve!

    Something I’ve noticed about the image is that it seems to have had its R and G values strongly boosted, I assume at the recording stage. At least you can adjust these using the levels window in Photoshop to turn the image colours of the dress deep black and blue and the pattern behind it black and white.

    https://www.facebook.com/photo.php?fbid=10152592552262063&set=a.10152407929897063.1073741825.664022062&type=1

    I wonder if the people who can see the dress as deep black and blue are to some extent making a similar adjustment in their visual system. I’m not one of them; instead I see the residual darkness and blueness of the dress area as illumination rather than local colour, and so read the dress as a white and gold object in a weakly bluish light. Many people who have seen the image both ways have described it online as a very dramatic difference, so for them at least there seems to be a change in their mode of perception involved rather than just a shift from one colour name to another.

    1. I agree David. I think one of the reasons why this phenomenon has been so interesting is that there is more than one factor at play. I think people can use different cognitive strategies to discount the effect of the illuminant.

Leave a Reply

Your email address will not be published. Required fields are marked *