Your Eyes Are Terrible At Seeing Blue

article by Ben Lincoln

If you've ever read about human vision, you're probably aware that your eyes are most sensitive to green, followed by red, and then blue. That is, if two objects reflect the same absolute amount of green and blue light, the green object will appear brighter because of your eyes' bias.

What you may not be aware of is how dramatic the effective difference between them is. Take a look at these two images, and make a note of the last coloured dot on each line that you can clearly discern the shape of. Don't strain your eyes or try to "improve" your results - this is an illustration of a physical fact, not a test of how great your vision is.

Coloured Dots

Constant Background

Gradient Background

Two variations on the theme of coloured dots. In the first, the dots are presented on a constant (black or white) background. In the second, the dots' background has an overall greyscale level that changes at the same rate as the colour level of the dots themselves.

The specifics of your results may vary dramatically depending on the quality of the display you view them on, as well as lighting conditions and other factors, but if your eyes are anything like mine, you'll notice these general observations:

In the top (white background) half of the first image, the yellow dots (fourth row) disappear significantly earlier than the others.
In the bottom (black background) half of the first image, the blue dots (fourth row) show a similar effect, and may not even be visible at all, even at the far right side of the image.
In the second image, a similar effect occurs, but concerning the first row of each half (the blue dots in the top half, and the yellow dots in the bottom half).

Pure yellow is the "anti-blue" - a mixture of red and green light - which is why you see a similar effect with that colour. I don't actually know why having a solid versus gradient background reverses the light/dark blue/yellow sensitivity - that was something I stumbled across when making the images for this article. If anyone has a good explanation, I'd love to hear it.

On a side note, if you can see any of the "stairstep" demarcations between shades of grey in the second image, congratulations: you disproved everyone who says "8 bits should be enough for everyone! Anything more than that is just marketing nonsense!"

This effect is not just a piece of trivia - it has important real-world implications. The single most dramatic example I've come across is in this image, which is from one of the test sets in the Colour Isolation Using Filters article.

Look at the first image. Do you see anything odd about the bananas?

How about in the second image, where I've explicitly highlighted the areas where the pattern is present?

Finally, examine the third image, which is a greyscale representation of the blue channel only.

A Real-World Example

Colour Image

Colour Image Highlighted

Greyscale Blue Channel Highlighted

Can you spot the pattern present in the blue channel of the first image? How about in the second image, where the areas of interest have been highlighted by green polygons? The final image shows only the blue channel (in greyscale form), where the pattern is easily seen. These images are a subset of one of the test image sets from the Colour Isolation Using Filters article.

Date Shot: 2011-06-24
Camera Body: Nikon D70 (Modified)
Lens: EL-Nikkor 80mm f/5.6 (metal-bodied) at f/11
Filters: Violet Stack, Omega 525DF20, Omega 670WB35
Date Processed: 2011-06-28
Version: 1.0

Like so many other things, our poor sensitivity to blue (and to a lesser extent, red) has both benefits and drawbacks. Taking advantage of it is part of the reason that Bayer-pattern filters can be used in digital imaging, and "lossy" image compression formats can discard considerable amounts of red and blue data before the quality of the image appears lower. On the other hand, it is easy to miss details "hiding" in these channels (particularly blue) when working with conventional colour or Three-Channel False Colour images.