Seeing in colour.

[ollypenrice]

There was a thread a while back disussing the strange way our eyes really see (or don't see!) in colour. On a recent visit AWR showed me this extraordinary demo of the effect. It illustrates the way that the colour we feel we are seeing in our peripherel vision is really just a fill-in from data gathered when our eyes had scanned the scene earlier. Only the central part of our view is really in 'live colour' if I have understood this correctly.

If you haven't seen this link then do take a look. Begin by scanning the image and then fix your eyes on the dot. Mouse over the picture without taking your eyes off the dot. You see a colour image. Then scan the picture away from the dot.

Uh-Oh!

It's amazing!

Olly

Big Spanish Castle

[brantuk]

Thats a curious effect - fascinating

[budski]

I (unknowingly) came across this effect a few years ago while working in Photoshop - I needed to see how colour pictures would look when going for magazine publication in black&white. I thought my monitor was playing up - but I couldn't replicate the effect. Don't we just love these biological mysteries......

[johnh]

The colour appearance is caused by the colour negative shock effect but it is amazing how long it lasts and movement kills it but it comes back but not as strongly eventually disappearing.

John.

[RogerTheDodger]

Another one, this flips automatically.

[Lynn]

that's amazing...rubbing eyes & watching again!!

[nightfisher]

wow that one works, so long as you keep your eyes focused on the dot, is persistance of vision involved

[umadog]

I don't think the illusion arises from previous scanning of the scene. You see colour throughout following the flip to gray-scale. There are small numbers of cones outside of your fovea. The difference between central and peripheral vision is not only colour, but acuity. Your acuity is much, much, poorer outside of the fovea. You don't notice this in everyday life because your eyes constantly scan the visual world using small movements known as saccades.

The illusion you show is thought to work roughly as follows. Colour information leaves the eye in three so-called "colour opponent" channels. You have red/cyan, blue/yellow, and also a light/dark ("black/white") pathway. Cone information is used for all three: the common misconception that rod vision does B&W is not correct.

So what does that have to do with the illusion? In the illusion you're initially staring at a bunch of colours. You adapt to those colours and when they switch you see the after-image (Afterimage - Wikipedia, the free encyclopedia). If you stare at magenta, you will see a green after-image. If you look carefully, you will see that the regions which appear green in the after-image are magenta before you mouse-over. Similarly, the sky is yellow in the initial image and this leads to the blue perception in the after-image.

There seems to be a second illusion taking place as well. The brightness in the initial colour image is the same throughout. The only information carrying the shape of the image (e.g. edges, etc) comes from colour. You will notice that the initial image looks indistinct. This is because the pathways which handle colour are bad at providing information on edges. When you mouse-over you see the B&W image. The pathways carrying information about shape use B&W contrast, so the B&W image looks high-resolution. When the illusion comes together the negative colour after-image is "painted over" the shape information carried by the B&W image on the screen. It is this which makes the illusion so compelling.

[Domis]

Wow,that is just too weird,I never heard this effect by the human eye before. Very interesting.

[umadog]

Pretty cool stuff, huh? The effect occurs mainly in the brain but has its roots in the way information is sent there by the eye. This effect is purely in the brain:

Akiyoshi's illusion pages

In fact, interesting point: the retina does a lot of processing of the image. Technically speaking, it's almost part of the brain. You can see the retina with the right equipment so that makes it the only part of the brain which is visible on a daily basis.

[ollypenrice]

I don't think the illusion arises from previous scanning of the scene. You see colour throughout following the flip to gray-scale. There are small numbers of cones outside of your fovea. The difference between central and peripheral vision is not only colour, but acuity. Your acuity is much, much, poorer outside of the fovea. You don't notice this in everyday life because your eyes constantly scan the visual world using small movements known as saccades.

The illusion you show is thought to work roughly as follows. Colour information leaves the eye in three so-called "colour opponent" channels. You have red/cyan, blue/yellow, and also a light/dark ("black/white") pathway. Cone information is used for all three: the common misconception that rod vision does B&W is not correct.

So what does that have to do with the illusion? In the illusion you're initially staring at a bunch of colours. You adapt to those colours and when they switch you see the after-image (Afterimage - Wikipedia, the free encyclopedia). If you stare at magenta, you will see a green after-image. If you look carefully, you will see that the regions which appear green in the after-image are magenta before you mouse-over. Similarly, the sky is yellow in the initial image and this leads to the blue perception in the after-image.

There seems to be a second illusion taking place as well. The brightness in the initial colour image is the same throughout. The only information carrying the shape of the image (e.g. edges, etc) comes from colour. You will notice that the initial image looks indistinct. This is because the pathways which handle colour are bad at providing information on edges. When you mouse-over you see the B&W image. The pathways carrying information about shape use B&W contrast, so the B&W image looks high-resolution. When the illusion comes together the negative colour after-image is "painted over" the shape information carried by the B&W image on the screen. It is this which makes the illusion so compelling.

Excellent, many thanks.

Olly

[AWR]

I don't think the illusion arises from previous scanning of the scene. You see colour throughout following the flip to gray-scale. There are small numbers of cones outside of your fovea. The difference between central and peripheral vision is not only colour, but acuity. Your acuity is much, much, poorer outside of the fovea. You don't notice this in everyday life because your eyes constantly scan the visual world using small movements known as saccades.

The illusion you show is thought to work roughly as follows. Colour information leaves the eye in three so-called "colour opponent" channels. You have red/cyan, blue/yellow, and also a light/dark ("black/white") pathway. Cone information is used for all three: the common misconception that rod vision does B&W is not correct.

So what does that have to do with the illusion? In the illusion you're initially staring at a bunch of colours. You adapt to those colours and when they switch you see the after-image (Afterimage - Wikipedia, the free encyclopedia). If you stare at magenta, you will see a green after-image. If you look carefully, you will see that the regions which appear green in the after-image are magenta before you mouse-over. Similarly, the sky is yellow in the initial image and this leads to the blue perception in the after-image.

There seems to be a second illusion taking place as well. The brightness in the initial colour image is the same throughout. The only information carrying the shape of the image (e.g. edges, etc) comes from colour. You will notice that the initial image looks indistinct. This is because the pathways which handle colour are bad at providing information on edges. When you mouse-over you see the B&W image. The pathways carrying information about shape use B&W contrast, so the B&W image looks high-resolution. When the illusion comes together the negative colour after-image is "painted over" the shape information carried by the B&W image on the screen. It is this which makes the illusion so compelling.

Thanks for this. As a colour-blind person, I've always been interested by this kind of illusion. It always seems strange that my brain flips the colours correctly when I cannot identify many of the colours in the original image.

Andrew

[umadog]

Exactly, so your colour channels are different. In what way they're different will depend on the sort of color-blindness you have. As you say, you will have the opponent channels.

It's instructive to think why we need multiple cone types for colour vision. Each cone type contains one pigment type. Each pigment type posses a characteristic light absorption spectrum (Cone cell - Wikipedia, the free encyclopedia). If we possessed only one pigment/cone type there would be a problem. An observer watching signals from that cone would see increasing and decreasing activity levels as the cone "sees" the changing visual world. However, they wouldn't know if these changing levels were due to variation in brightness, variation in light wavelength, or both. Either could cause the signal to change. Adding a second cone/pigment type allows the observer to disassociate wavelength from intensity. Basically, information about wavelength can now be derived from the ratio of the two cone activities. Information about intensity from the sum of the two cone activities.

So why have three cone types? It turns out that more than one wavelength will produce the same activity ratio. Which means these wavelengths will all look the same. Hence the term "colorblindness." Adding a 3rd cone type decreases the number of such wavelength confusions. That's why someone with three cone types can distinguish more colours than someone with two cone types.

Most mammals have two cone types. So dogs and cats are colour-blind in the same sense as some people are colour-blind. They don't see black and white, as is commonly thought, but a more restricted colour set. Fish and reptiles generally have 4 cone types, so they have better colour acuity than we do. Birds also have 4 but their photoreceptor cells can also posses coloured oil droplets which filter the incoming light before it reaches the photo-sensitive protein. This means they effectively have more then 4 cone types. Cool huh?

[umadog]

Here's the bird stuff: Bird vision - Wikipedia, the free encyclopedia

See the cone pigment curves? They can see into the near-UV.

[ollypenrice]

Better and better!!!

Olly

[DaveGarland]

Thats really cool! I love things like this!