Tennis ball color theory

[rutt]

Ginger and I have had little friction about the color of tennis balls. I think I can state her side fairly as follows:

The dog in my avatar has a green tennis ball in her mouth. Tennis balls are yellow and not green.

(Correct me if I am wrong, Ginger.)

While, I grant that Ginger is an expert on tennis balls, my position on this has been:

After my labrador retriever has owned a tennis ball for a while, it does turn green.

I''ve even offered to send Ginger some of these green tennis balls so she can verify this for herself, but she never took me up on this offer.

I decided to use applied color theory to settle this dispute once and for all. The results surprised me and should settle the issue once and for all.

Here is a picture taken in overcast daylight which shows a tennis ball that has been in the position of my dog (left) and a new one stolen from my wife's bag (right).



This shows that my dog really can make a dramatic impact on tennis ball coloration. The ball on the right is certainly yellow and the ball on the right looks dark green. But we don't have to trust our eyes on this. We can measure the color values of both balls using photoshop color samplers:





The color values for each ball are:

1. Dog's ball: R=141, G=136, B=78, C=48, M=42, Y=80, K=10
2. New ball: R=201, G=223, B=123, C=25, M=2, Y=60, k=0

I was expecting to find that the dog's ball would have more cyan and roughly equal amounts of magenta and yellow as the new ball, making it more green. I was expecting to find that the dog's ball would have a higher green value relative to its red and blue values compared to the new ball. But this isn't what the numbers show.

The numbers show that the dog's ball is just plain darker than the new ball. I suppose that isn't really surprising, given what she does with it.

So why is this interesting? I think it shows that we don't perceive "dark yellow" as yellow, but rather as a dark shade of green. Which I suppose makes it green, after all!

So, Ginger, while I greatly respect your knowledge of tennis balls and the color of tennis balls that haven't been in the possession of labrador retrievers, I think I have shown that there are tennis balls that appear green and have explained why.

[pathfinder]

Originally Posted by rutt

Ginger and I have had little friction about the color of tennis balls. I think I can state her side fairly as follows:

The dog in my avatar has a green tennis ball in her mouth. Tennis balls are yellow and not green.

(Correct me if I am wrong, Ginger.)


While, I grant that Ginger is an expert on tennis balls, my position on this has been:

After my labrador retriever has owned a tennis ball for a while, it does turn green.

I''ve even offered to send Ginger some of these green tennis balls so she can verify this for herself, but she never took me up on this offer.


I decided to use applied color theory to settle this dispute once and for all. The results surprised me and should settle the issue once and for all.

Here is a picture taken in overcast daylight which shows a tennis ball that has been in the position of my dog (left) and a new one stolen from my wife's bag (right).



This shows that my dog really can make a dramatic impact on tennis ball coloration. The ball on the right is certainly yellow and the ball on the right looks dark green. But we don't have to trust our eyes on this. We can measure the color values of both balls using photoshop color samplers:





The color values for each ball are:

1. Dog's ball: R=141, G=136, B=78, C=48, M=42, Y=80, K=10
2. New ball: R=201, G=223, B=123, C=25, M=2, Y=60, k=0

I was expecting to find that the dog's ball would have more cyan and roughly equal amounts of magenta and yellow as the new ball, making it more green. I was expecting to find that the dog's ball would have a higher green value relative to its red and blue values compared to the new ball. But this isn't what the numbers show.

The numbers show that the dog's ball is just plain darker than the new ball. I suppose that isn't really surprising, given what she does with it.

So why is this interesting? I think it shows that we don't perceive "dark yellow" as yellow, but rather as a dark shade of green. Which I suppose makes it green, after all!

So, Ginger, while I greatly respect your knowledge of tennis balls and the color of tennis balls that haven't been in the possession of labrador retrievers, I think I have shown that there are tennis balls that appear green and have explained why.

Now this is truly funny!!! And educational to boot. Great post John, I am sure Ginger will be pleased also.

[ginger_55]

Rutt, I am flattered and scientifically excited. You do realize that you could make a scientific paper out of those facts, have it published, and so your new life would begin.

In my sojourn into photographing tennis balls for a Challenge, I did discover that it is very difficult not to photograph a green tennis ball. Upon trying to correct for that, I discovered that the D balls had about every color in the spectrum in them, as you discovered. A simple yellow tennis ball is NOT. Simple yellow.
Things are often not as they seem. And other cliches. Cliches, of course, are born out of truths.

I never posted re my tennis ball discovery, I was too busy trying to figure out what bee had gotten into the tennis ball industry's bonnet.

I got one shot I could show as a true yellow, after some mods. (Curves or levels )

ginger

[wxwax]

I always thought the color was a dayglo lime green, same as some safety vests. A color not found in nature, as they say, and because of that it stands out more to the human eye.

[DJ-S1]

The color is listed as "optic yellow". Don't know what that means exactly, but a web search shows tons of offers for glasses/lenses that mute most colors except optic yellow. Supposed to make the ball really jump out at you! Do you use them, Ginger?

[rutt]

I missed an important point. It turns out that the color sampler measurements in LAB are very interesting:

1. Dog ball: L=69, A=-8, B=28
2. New ball: L=91, A=-15, B=41

These numbers show conclusively that the new ball is not only brighter than the dog ball (L) but also both more green (A is more negative) and more yellow (B is more positive). So Sid's assertion that new tennis balls are both green and yellow is verified. The new ball is 31% brighter than the dog ball, about 91% more green, and 46% more yellow. So the compared to the new ball the dog ball is, if anything, more yellow green. Yet, because it is darker, we perceive it to be more green.

We can conclude, once again, that our brains process color information in complex and very nonobvious ways.

[mercphoto]

Originally Posted by rutt

We can conclude, once again, that our brains process color information in complex and very nonobvious ways.

Both vision and hearing are complicated. Have you ever done a blind spot test? Very interesting, and shows you that the brain not only processes information, but sometimes ADDS information. Also, video compression for both pictures and movies takes advantage of how your brain processes vision so as to remove stuff that your brain won't notice, or will correct for. Hearing has equally interesting strangeness to it. :)

[wxwax]

Cool stuff, Merc.

Rutt, thanks for this very interesting thread. I like the way you broke down the color information. It's edumacational.

[rutt]

Originally Posted by Molsondog

Now I know why I watch this forum. There is always something to learn. Back in grad school, our color theory professor never made learning this much fun. However, that was in the days of slide rules and no Photoshop. Thanks for the day brightener. It's only appropriate that the values are LAB(rador) values! I've attached a photo of what's really on a dog's mind. It's a repeat, but fits.

I saw this before and loved it then. An oldie, but one of the greats.

[wxwax]

First time I've seen it.

[Baldy]

Great post, Rutt!

Some people say the tennis ball in our calibration print looks a little yellow and that tennis balls should be a more vibrant shade of green. Do we know if tennis balls are a little bit out of gamut in sRGB, the color space of the web?

[al'be:do]

Hey, Rutt!
This is a great and funny idea of applying color science to every day objects!

Sorry for bumping into this thread, but I think I can tell something that might be interesting for some people.
The thread is old, but I found it via google while searching something about fluorescent spectra, so the information in it might still be interesting to others.

Btw, this is my first post here. :)

Originally Posted by Baldy

Great post, Rutt!

Some people say the tennis ball in our calibration print looks a little yellow and that tennis balls should be a more vibrant shade of green. Do we know if tennis balls are a little bit out of gamut in sRGB, the color space of the web?

Tennis balls are not only out of the sRGB gamut but also out of the human color gamut for body colors - thus also out of CIELAB aka Lab. This is because it's a fluorescent color. The fluorescent yellow with the name optic yellow seemingly reflects more light than it receives.
A non-fluorescent color with the same chromaticity (a*,b*-coordinates) can't be as bright as the fluorescent yellow of a tennis ball because the energy of the reflected light is limited to the energy of the illuminating light source.

The trick of fluorescent colors is that parts of the ultraviolet light - which is invisible to us - are transformed into visible light. High energy uv photons excite electrons which emit lower wavelength (visible) light when jumping back to their ground state. The excess energy is radiated as heat.

Because we don't see UV light we have the impression that the ball reflects more light than what we expect from a ball of this color. This lets the ball appear to "glow" or look like being a self luminous light source.

If you look at a gamut in 3d, for example the CIELAB/Lab gamut then you have absolute saturation limits for every chromaticity coordinate - the MacAdam limits. The surface of the gamut shows which reflective or transmissive surface colors are theoretically possible. Fluorescent colors and self luminous light sources can exist beyond the MacAdam limits. This is the case for the fluorescent yellow color of tennis balls.

In the case of the chewed tennis ball there's another reason for the color shift - which is the main reason in my eyes:
All the dirt and uv radiation this ball has to suffer in it's life degrades it's ability to receive or transform uv light. The dirt works as a uv filter and permanent uv radiation destroys the fibres and maybe even the fluorescent pigments.
If the tennis ball cannot receive uv radiation anymore then the fluorescent mechanism is dead. So the ball looks like being of an ordinary color and thus a lot less bright and of course has a different color because the portion that was re-radiated from the uv input is missing. The other obvious effect of dirt is that also visible light is filtered, which also makes the ball appear even darker.

Color analysis with the eyedropper is fun but not always helpful or reliable. To see what really happens you would need a spectral analysis with a densiometer or a color analysis with a colorimeter. Using a photo for analyzing color is - as you might know - not very reliable and cameras aren't the most reliable instruments. ;) As soon as a color is outside the camera gamut every eyedropper measurement is useless.

Another reason for chromaticity shifts is the chromaticity rotation that depends on the lightness and saturation of the color in the CIELAB space. CIELAB isn't perceptually very uniform - especially in the green and blue region. It's still used as standard although there are better and newer color spaces available, so be cautious and always check if colors in te measuring region tend to shift simply because of the nonuniformity of the color space itself.
Daytime fluorescent yellow-green (for traffic signals) has a defined color region between the xy coordinates (0.387,0.610), (0.369,0.546), (0.428,0.496), (0.460,0.540). For a flourescent color to appear flourescent in daylight the MacAdam limit luminance factor YM has to be over 70% of the maximum possible lightness of the according optimal color (Macadam limit), so we take the lowest limit for fluorescence: Y=0.7.

Now I take the third of those four coordinates (the "inner" boundary that's less saturated) and transform the coordinate into Lab (without the proper illuminant adaptation, as the importance of color adaptation is only secondary in our case):

Yxy (0.7,0.428,0.496) = Lab (128.99,-22.70,107.36)

Compare it to the values in the photo: (91,-15,41) You can see that the photo shows the colors a lot less saturated than they actually are.

We see that
1. The lightness is over 100! So it can't be coded in Lab. Edit: Sorry for this error, the lightness isn't over L=100, but it's certainly higher than the according maximum lightness of this color in the Lab color space. I'll recalculate the value and correct this later.
2. the b*-coordinate can be over 127 (up to 200 if you take the first of the four coordinates, for example) for fluorescent yellow. So in some cases the color might not be in the defined region for Lab colors in Photoshop.

Every measurement you make with an eyedropper is actually just for fun but rather useless in this case. The new ball's colors appear less saturated than in the real world, due to gamut restrictions of the camera. The other ball's colors are inside the gamut. So color shifts due to the gamut compression of the camera render the results useless.

Baldy, the tennis ball appears only yellow in your test image because it's lightness doesn't fulfill the criteria for fluorescence. Because fluorescent objects appear brighter than normal the hunt effect can also come into play, which makes brighter colors appear more saturated.
The other reason is that there are several dyes for fluorescent yellow with slightly different chromaticities (see above for the color range for daylight fluorescent yellow).