Purple Flowers -> blue

[ChrisDauer]

I was looking over images recently and I noticed that a few of my shots from my Pentax Optio S5z showed images of flowers that were purple, as blue.

When I compared similiar shots from my Canon 5D, they were indeed, purple. Maybe not the exact shade, but a whole lot closer than the blue that the Pentax was giving me.

My question is why/how this occurs and what can I do about it?

 

[Bob]

Purple s a really tough color.
First it exists nowhere but in the mind of the beholder. You will not find it in the spectrum at all.
Instead, our brains tell us that we see purple when it is stimulated by a combination of red and blue wavelengths. Not that there exists significant individual variability in the necessary mix of reds and blues to achieve a "purple" stimulus. So, there is probably no real solution to this unless the complete color reproduction system is capable of replicating the full spectrum and the mixture of wavlengths that occur in nature. This is something that is just NOT POSSIBLE from systems that use combinations of red blue and green to fake out all visible color.
-bob

 

[ChrisDauer]

Come on Bob... Violet.

ROY G. BIV from High School Physics. Heck, I'll take Indigo if need be. And clearly the FLOWER is purple

What I don't get on the visible light spectrum; is if the sensor can fake Yellow by a mixture of Green and Blue, then why can't it fake Purple by a mixture of Red and Blue? Why is Purple more difficult than Yellow? Is it because the sensor has twice as many Green receptors?

I just want my sensor to duplicate what I see in RL. The Canon is off a shade, but at least it falls in the right category. I mean, geeze, Leica has mastery on putting purple into the image! See my 50/1.0 image posting. :ROTFL:

 

[TRSmith]

Maybe this is a dumb question, but if there are options for the camera's color space, are they both set to the same space? For instance, Adobe RGB vs. sRGB?

 

[Bob]

Come on Bob... Violet.

ROY G. BIV from High School Physics. Heck, I'll take Indigo if need be. And clearly the FLOWER is purple

What I don't get on the visible light spectrum; is if the sensor can fake Yellow by a mixture of Green and Blue, then why can't it fake Purple by a mixture of Red and Blue? Why is Purple more difficult than Yellow? Is it because the sensor has twice as many Green receptors?

I just want my sensor to duplicate what I see in RL. The Canon is off a shade, but at least it falls in the right category. I mean, geeze, Leica has mastery on putting purple into the image! See my 50/1.0 image posting. :ROTFL:

As the wavelengths get shorter they get less red (which is longer)
ROY G BIV contains purple nowhere. That is an utterly ridiculous mnemonic, similar to BBROYGBVGW which I will not describe in mixed company. BBROYGBVGW on the other hand accurately describes the IEC and ANSI color codes, but does not really describe the spectrum, ROYGBIV contains no purple at all. Go ahead and take a prism and try it out. Or on the other hand, you could tell me what IS the wavelength of purple. All sorts of physicists have been waiting to know.
-bob
-bob

 

[sizifo]

Maybe this will help http://www.madsci.org/posts/archives/2001-03/985572799.Ph.r.html with the physics of purple. No, there is no "wavelength for purple", but from the explanation in the link it would seem that rgb colorspace should be well suited for reproducing purple, since the receptors in the eye are tuned to red green and blue. From the discussion here I haven't yet been able to understand why purple should be so difficult to reproduce.

 

[nostatic]

so why is there no blue food?

 

[Jack]

While there is no purple in the visible spectrum, our eyes capture light in basically an RGB system of cones -- so we see a color we call purple anytime there is an absence of green...

So yes, RGB digital devices are capable of replicating it and obviously some are better than others at doing it than others. I would suspect this posibly has to do with where the red and blue spectrums get cut off by any internal IR or UV cut filters...

Cheers,

 

[Robert Campbell]

I seem to remember that there are seven colours in the visible spectrum because Newton thought that there was something special about the number 7, and forced 7 colours into the spectrum - something alchemic, perhaps.

Blue flowers can be difficult to reproduce well: but surely this is because of ultra violet radiation which we can't see, but which sensors and film are sensitive to: so perhaps the film/ sensor is recording a greater range of wavelengths - so the apparent colour is more 'accurate'. If you were an insect, being able to see beyond the range of wavelengths visible to us is useful; and if you were an eagle, seeing the uv trail of your prey's urine might give you a tasty dinner...

 

[Robert Campbell]

so why is there no blue food?

There once was a range of 'water ices' with colours which didn't represent the taste presented; the blue colour was, if I remember correctly, strawberry. When our kids visited the blue grotto in the Rhone glacier, one of them licked the ice, and said it tasted of strawberry...

Lobsters are blue, at least before they are cooked.

Black coloured animals or birds don't taste good: this acts as a warning to predators, so perhaps the same is true for anything blue.

 

[TRSmith]

Musn't forget the wonderful (and native to this great State) Blueberry.

 

[jonoslack]

I was looking over images recently and I noticed that a few of my shots from my Pentax Optio S5z showed images of flowers that were purple, as blue.

When I compared similiar shots from my Canon 5D, they were indeed, purple. Maybe not the exact shade, but a whole lot closer than the blue that the Pentax was giving me.

My question is why/how this occurs and what can I do about it?

Hi Chris
I don't pretend to understand this (my techie credibility is very low), but we have some pansies in the kitchen I've been photographing, with the same problem.
Still, film is, if anything, worse - I remember bluebell woods being the biggest catastrophe.

The only consolation I can offer is that blue is a fine colour, and it's unlikely that your client/viewer will notice the difference:grin:

 

[jonoslack]

This is what the Ricoh GX-100 does

Very nice, but these should certainly be purple - the top edge of the right hand flower get's close to it . . but not close enough.

No trouble with tomatoes though :grin:

 

[ChrisDauer]

The only consolation I can offer is that blue is a fine colour, and it's unlikely that your client/viewer will notice the difference:grin:

But I'm my client/viewer!
And it erks me to see (or not as is the case). :angry:

 

[nostatic]

Musn't forget the wonderful (and native to this great State) Blueberry.

blue on the vine, purple on the plate (or should I say, "absence of green" on the plate ).

 

[LFPhoto]

Purple is a hard color for many sensors - others have no problem. If your sensor has a problem two suggestions: First is to manually set white balance, second is to try a hot mirror filter. Works on the D2H.

Brian

 

[scott kirkpatrick]

Re: the color Purple

I think blues and purples are at the edges of the gamut that s-RGB is supposed to render, and only a little inside the a-RGB space. And if you are viewing a jpg in a browser, your onscreen impression is very hard to predict. I took two test shots tonight in difficult conditions. We don't have anything flowering with the right colors so I grabbed a grey-blue-lavender polypropylene sweater, a purple sponge, some orange and red fruit, a bright red cactus and a reddish purple box of Kleenex, shot it under dimmed halogen illumination, of about 2000 - 2500K. I figured this would give any camera indigestion.

The first shot, with an M8, 35mm shot at f/2.8, ISO 160 (UV/IR filter and lens detection ON) came up with an "as shot" color temperature of 2940K and shift of -2 (toward the green). Clicking on the WhiBal card brought that down to 2530K and -1. This is typical of the M8, since the new AWB firmware strenuously resists going below 3000K. The engineers felt that if you take pictures by candlelight, you want to get a warm overall coloration.

The second shot, with a GRD-II, came up in C1-4.01 with 2450K +11 (C1's profile for this camera is a real generic neutral translation, without camera-specific tuning), and click-balanced to 2215K +12. ISO 100, f/2.8.
The results (the M8 shot is pretty accurate, the GRD-II is way off to the blue):

 

[scott kirkpatrick]

Here's the closest I can come to the sweater's dull purple with the GRD-II (by tweaking the white balance to about 2600K +10). I still don't have an orange orange.

 

[Terry]

Yikes on the Lemon whoops orange!

 

[Bob]

Purple has been tough from the beginning of my career. My first non-medical job was to create fabric color combinations that would faithfully print true in the fabric mills. We had a limited number of non-fugitive dyes available to us and not too many economical purples. The real problem with purple is the way that the eye sees it in varying illuminants. There has to be enough red and enough blue, but a purple dress has to still be purple under candle light or it is a failure. What I am talking about is the physiological sensation of purple. Jack mentioned that purple is the absence of green. That is almost right, since magenta is sort of purple, but it is not all that purple has to offer. Dyes absorb light of varying wavelengths to give us the apparent color we see. There is nothing magic about red blue and green other than they are convenient. The red blue and green sensitivities of human comes do not may that closely to the red green and blue color filters that are used in out digital sensors. (see http://en.wikipedia.org/wiki/Color_vision). If a deficit of green occurs in the illuminant spectrum, the ability to generate a credible physiological purple in any dye or pigment combination is handicapped unless other colors are re-mapped according to the Ives transform q.v.
All that said, really, and in summary, creating or duplicating a color that does not exist as a spectral color in a true sense, but does exist in our mind's eye, and keeping it relatively true under a variety of illuminants, is basicly a bitch.
So, I think that there is no "true" rendering in the srgb space or even the prophoto space of a color that is quite like purple madder or purple irises. I have yet to see a photograph of the object that comes close to the real pigment as the light changes.
In order to do that, one would have to duplicate the reflection spectrum of the original, which cannot be universally be done with red blue and green.
-bob (back in Massachusetts)