Do camera companies actually *want* accurate color in their cameras?

A thought occurred to me after watching Hudson Henry's camera calibration tutorial. In it, he shows how to use a simple color checker tool to ensure that your colors are accurate. The procedure is quite simple, basically you use a standard reference to profile your camera and then your photo editing software makes adjustments to account for any variation.

But it seems to me that camera companies must have already heard about this technology. So it leads me to one of the following thoughts:

a) camera companies do not have the technology to build an ICC profile for their particular sensor and only X-rite (and Datacolor) have this technology. I find this unlikely.

b) camera companies do understand the need for calibrating their cameras, but because of variation from camera to camera, it has to be done individually for each unit, and that is not cost-effective for the camera company, so they leave it to you to do it if its important. That would suggest that there is visible difference between one Nikon D5 and another Nikon D5

c) the color variance is so small that its not worth doing, and that color profiling is only necessary for very few people (but we can all clearly see the variation with our own eyes, so again I think this unlikely)

d) they actually do not wish to have accurate color, for whatever artistic/marketing reasons, and if you want accurate color, you will have to do it yourself. This would also imply that for the majority of photographers, accurate color may not be the ideal choice.

Or is there something else that I'm missing? And yes I realize there is a lot of oversimplification here, like I realize that RAW files have no color its up to the RAW processor to interpret the color, but there must be some variance even then because otherwise RAW processors would simply create a profile for each camera (I mean they do it already) and there would be no need for calibration at all.

d)

See, that wasn't so hard.

Kelly

You have overlooked a crucial element. A camera cannot accurately reproduce the colour of a scene without knowing the colour of the light illuminating that scene. The colour of that light is a huge variable and is why cameras cannot be accurately calibrated once and for all.

Even in bright sunlight, the colour of that light varies significantly with (i) the atmosphere, which absorbs and scatters some of the sunlight; and (ii) with surrounding coloured objects that reflect light and produce colour casts on the subject of the picture. For example, a person standing in bright sun near some trees may get a slightly green colour cast because of green light reflected from the leaves of the trees.

MinAZ said:

A thought occurred to me after watching Hudson Henry's camera calibration tutorial. In it, he shows how to use a simple color checker tool to ensure that your colors are accurate. The procedure is quite simple, basically you use a standard reference to profile your camera and then your photo editing software makes adjustments to account for any variation.

But it seems to me that camera companies must have already heard about this technology. So it leads me to one of the following thoughts:

a) camera companies do not have the technology to build an ICC profile for their particular sensor and only X-rite (and Datacolor) have this technology. I find this unlikely.

b) camera companies do understand the need for calibrating their cameras, but because of variation from camera to camera, it has to be done individually for each unit, and that is not cost-effective for the camera company, so they leave it to you to do it if its important. That would suggest that there is visible difference between one Nikon D5 and another Nikon D5

c) the color variance is so small that its not worth doing, and that color profiling is only necessary for very few people (but we can all clearly see the variation with our own eyes, so again I think this unlikely)

d) they actually do not wish to have accurate color, for whatever artistic/marketing reasons, and if you want accurate color, you will have to do it yourself. This would also imply that for the majority of photographers, accurate color may not be the ideal choice.

Or is there something else that I'm missing? And yes I realize there is a lot of oversimplification here, like I realize that RAW files have no color its up to the RAW processor to interpret the color, but there must be some variance even then because otherwise RAW processors would simply create a profile for each camera (I mean they do it already) and there would be no need for calibration at all.

Click to expand...

I think the answer is hanging somewhere around (d). The market doesn't want accurate colour, it wants accentuated colour. Even in the days of colour film this was true, and films which offered very true to life colour tended not to last long.

A couple of additional points.

Whilst a colour checker may be useful for scientifically accurate colours, what matters to most of us is perception. For example, if you step into bright sunlight from a dark room the scene will appear to you quite different from the way it will appear after an hour or so. Which would be accurate?

The effect of the colour of the light on the scene has already been mentioned, but the same applies if you are looking at a print under different lighting conditions. This is obviously more difficult, since a print on the wall will have different light during the day, on different days and at night, for example.

So for me at least, the search for accurate light is chasing shadows (no pun intended!), and I'll aim for what is most aesthetically pleasing (to me!)

They bring the camera's colors right up to a threshold then stop. this would allow you to edit, enhance these colors during post processing. so the answer is no.

IMHO, you overlook a few aspects of colour rendition that makes the question almost impossible to answer.

The natural light is a mix of light of different wavelengths. We do not sense the wavelengths but the mix of how the three types of receptors in our eyes are stimulated. There's some heavy processing going on in our brain even before a camera comes in the path.

You are right that most sensors only see light levels, not colour, and that the colours are recreated by an algorithm. The dyes in the Bayer filter can probably change slightly between production runs and may deteriorate slightly over time in a non-linear and unpredictable way.

Many modern light sources are not black body radiators giving off a continuous spectrum. Both flourescents and LEDs are known to produce spikes and valleys in their output. Some of these irregularities may interfere with the Bayer filter colours in unexpected ways, but differently with every type and make of light source and with different Bayer filter dyes, so to be absolutely precise, a calibration must take the spectrum characteristics of every light source into account.

You are probably right in assuming that accurate colour is not necessarily the most desired quality. On my camera and in my raw converter, I can select any number of colour scemes, like Vivid, Neutral, Portrait, Standard, etc. and I'm sure the manufacturers are going for pleasing rather than accurate.

I'm just an amateur photographer, so what interests me is the ability to create a reasonably convincing picture that resembles what I saw and will resonate with the people who see it. For my needs, accuracy of colour is more nice-to-have than need-to-have. If I were selling clothes on the Internet, I'd probably see it differently.

I often wonder how colour can survive at all in the chain starting with crappy artificial light over subjects painted or made up with artificial dyes, through three-colour Bayer filters, algorithmic conversion, jpeg viewers, uncalibrated screens of uncertain characteristics to our eyes where the whole picture is once again filtered into a combination of three colour channels before it is processed into a perception.

Good luck and good light.

MinAZ said:

A thought occurred to me after watching Hudson Henry's camera calibration tutorial. In it, he shows how to use a simple color checker tool to ensure that your colors are accurate. The procedure is quite simple, basically you use a standard reference to profile your camera and then your photo editing software makes adjustments to account for any variation.

But it seems to me that camera companies must have already heard about this technology. So it leads me to one of the following thoughts:

a) camera companies do not have the technology to build an ICC profile for their particular sensor and only X-rite (and Datacolor) have this technology. I find this unlikely.

b) camera companies do understand the need for calibrating their cameras, but because of variation from camera to camera, it has to be done individually for each unit, and that is not cost-effective for the camera company, so they leave it to you to do it if its important. That would suggest that there is visible difference between one Nikon D5 and another Nikon D5

c) the color variance is so small that its not worth doing, and that color profiling is only necessary for very few people (but we can all clearly see the variation with our own eyes, so again I think this unlikely)

d) they actually do not wish to have accurate color, for whatever artistic/marketing reasons, and if you want accurate color, you will have to do it yourself. This would also imply that for the majority of photographers, accurate color may not be the ideal choice.

Or is there something else that I'm missing? And yes I realize there is a lot of oversimplification here, like I realize that RAW files have no color its up to the RAW processor to interpret the color, but there must be some variance even then because otherwise RAW processors would simply create a profile for each camera (I mean they do it already) and there would be no need for calibration at all.

Click to expand...

sure

www.flickr.com/photos/mmirrorless

Tom Axford said:

You have overlooked a crucial element. A camera cannot accurately reproduce the colour of a scene without knowing the colour of the light illuminating that scene. The colour of that light is a huge variable and is why cameras cannot be accurately calibrated once and for all.

Even in bright sunlight, the colour of that light varies significantly with (i) the atmosphere, which absorbs and scatters some of the sunlight; and (ii) with surrounding coloured objects that reflect light and produce colour casts on the subject of the picture. For example, a person standing in bright sun near some trees may get a slightly green colour cast because of green light reflected from the leaves of the trees.

Click to expand...

I think you may be confusing camera calibration with white balance. This sort of calibration is usually done in a controlled situation under daylight balanced light, but sometimes also done under varying conditions to create different profiles. After that, a separate test can be done to check for white balance (and I'm not even entirely sure color cast can be corrected for easily tbh).

Klaus dk said:

...

I often wonder how colour can survive at all in the chain starting with crappy artificial light over subjects painted or made up with artificial dyes, through three-colour Bayer filters, algorithmic conversion, jpeg viewers, uncalibrated screens of uncertain characteristics to our eyes where the whole picture is once again filtered into a combination of three colour channels before it is processed into a perception.

Click to expand...

+1

Or printed on printers that have their own characteristics.

This is something that I have to deal with with paid jobs.

It is expected that I produce pleasing pictures.

For that I use X-Rite equipment.

When I shoot portraits, I'll use an X-Rite Colorchecker Passport as a target.

And I have my monitor calibrated with an X-Rite Color Munki Display.

And I have the printer profile from the commercial printer I submit pictures to installed in Lightroom so I can do a soft proof to simulate what this image would look like printed on their printer.

So . . .

1. I take a picture of the X-Rite Colorchecker Passport while doing the portait session.

2. I load that picture up in Lightroom. Export as DNG. Use the X-Rite software to create a profile. Load that back in to LR. And then apply that profile to the shots from that session.

3. I then edit the pictures in LR on my colour calibrated monitor, that I calibrated with X-Rite Color Munki Display.

4. And then I preview what it would look like printed with the printer profile I received from the commercial printer I use, that has been loaded into my copy of LR.

I may make tweaks based on personal preference.

But, using this equipment gets me to the starting line really quickly!

Take care & Happy Shooting!

Klaus dk said:

Good luck and good light.

Click to expand...

MinAZ said:

Tom Axford said:

You have overlooked a crucial element. A camera cannot accurately reproduce the colour of a scene without knowing the colour of the light illuminating that scene. The colour of that light is a huge variable and is why cameras cannot be accurately calibrated once and for all.

Even in bright sunlight, the colour of that light varies significantly with (i) the atmosphere, which absorbs and scatters some of the sunlight; and (ii) with surrounding coloured objects that reflect light and produce colour casts on the subject of the picture. For example, a person standing in bright sun near some trees may get a slightly green colour cast because of green light reflected from the leaves of the trees.

Click to expand...

I think you may be confusing camera calibration with white balance. This sort of calibration is usually done in a controlled situation under daylight balanced light, but sometimes also done under varying conditions to create different profiles. After that, a separate test can be done to check for white balance (and I'm not even entirely sure color cast can be corrected for easily tbh).

Click to expand...

I am assuming that by using the X-Rite, you can do both at once.

Since the colours on the X-Rite Color Checker Passport are known values, the software can create a profile that shifts colours around so that the resulting image will be close to the original values of the target.

As for white balance, you could just take a picture of the X-Rite target in the light you are shooting in and the profile will be built to remove the colour shift.

When I first got my X-Rite Color Checker Passport, I took a test shot in my studio with an orange gel put over my flash unit.

When I created the profile from that target shot, shot with orange light, and I applied the profile to the test shot, it removed the orange colour that was introduced with the orange gel. All I had to do was use the WB tool to click on one of the available "whites" on the X-Rite Color Checker Passport to decide if I wanted a warm, cool or neutral white.

I found the test shot.


This test shot was done with a CTO (orange) gel over the flash unit. X-Rite Color Checker Passport made it quick and easy to have a well balanced picture.

As a reference, here is the shot without the profile and not selecting WB from the target.


This is the test shot without the profile or WB applied. NOTE: There is a CTO (orange) gel over the flash unit. That is why the shot is orange. LOL. I just wanted to see how well the X-Rite Color Checker Passport worked at getting me to neutral colours quickly and easily.

Take care & Happy Shooting!


--
My Personal Flickr Favs . . .
http://www.flickr.com/photos/tacticdesigns/sets/72157631300869284/
[FL][RP][LS][GC][51][ML][TMPM][ExifTool]

Tom Axford said:

You have overlooked a crucial element. A camera cannot accurately reproduce the colour of a scene without knowing the colour of the light illuminating that scene. The colour of that light is a huge variable and is why cameras cannot be accurately calibrated once and for all.

Even in bright sunlight, the colour of that light varies significantly with (i) the atmosphere, which absorbs and scatters some of the sunlight; and (ii) with surrounding coloured objects that reflect light and produce colour casts on the subject of the picture. For example, a person standing in bright sun near some trees may get a slightly green colour cast because of green light reflected from the leaves of the trees.

Click to expand...

e) What Tom said.

Regarding jpg captures, camera companies make cameras to try to have the best colors possible in auto white balance mode but they also provide other white balance presets like daylight, cloudy, tungsten, fluorescent, etc. for other common light sources. And they provide custom white balance settings where the user can set white balance according to the light sources at the specific location.

The user must be intelligent enough (and not lazy) to take advantage of the capabilities of the camera and tweak in post processing.

Raw of course depends on the software and user's expertise.

And yes, IMO, camera companies do actually *want* accurate colors in their cameras.

MinAZ said:

...

Or is there something else that I'm missing? And yes I realize there is a lot of oversimplification here, like I realize that RAW files have no color its up to the RAW processor to interpret the color, but there must be some variance even then because otherwise RAW processors would simply create a profile for each camera (I mean they do it already) and there would be no need for calibration at all.

Click to expand...

IDK.

The way I think of it is . . . the colour of light is going to be different. (As pointed out by other posters.)

Cameras are getting better at guessing at the colour of light. Is it daylight? is it tungsten? Is it fluorescent?

But what if it's a mixture?

Unless you have someone walking around in the scene with an X-Rite Color Checker Passport stuck to their chest . . . (or better yet, an X-Rite Color Checker T-Shirt!!!), your guess and the camera's guess is just that . . . a guess.

If there is something white in the scene, you could do a WB off of it.

But who's to say that the light you are taking a picture with isn't devoid in a particular band of light?

I don't really see how calibrating a camera beyond what they are already calibrated can help out.

For a complete solution, from what I am seeing, you have to take the colour of the light into account.

If so, something like the X-Rite Color Checker Passport target is an easy solution.

You take a picture of the target in the light that you are shooting in.

It can create a profile that takes into account your camera.

And when you click on the different available "white" targets, you can white balance your picture easily.

To me, it seems like the simplest solution, and not something the camera manufacturers should really worry about, unless they want to come up with their own targets.

Take care & Happy Shooting!


--
My Personal Flickr Favs . . .
http://www.flickr.com/photos/tacticdesigns/sets/72157631300869284/
[FL][RP][LS][GC][51][ML][TMPM][ExifTool]

I think the idea of accurate color is flawed. To accurately access the color you need a spectrometer.

Unless the spectral response of the red,green & blue channels of the output device exactly match those of the camera it is impossible to get accurate color for all viewers.

The human eye is not a standard device which works the same for everyone. Some people have some sort of color blindness so that specific colors are indistinguishable, then you have variations in how far into the UV & IR ends of the spectrum people can see...

Cameras are set to be able to give reasonably accurate colors to the average human, they do quite well at this, but wouldn't stand a chance of fooling other creatures which often see more colors than we do - A mantis shrimp apparently has 13 types of color receptor & IIRC a horse has 5...

I think you might have missed that the color checker is supposed to be used for each different light source you shoot in. It isn't a one off calibration.

MinAZ said:

A thought occurred to me after watching Hudson Henry's camera calibration tutorial. In it, he shows how to use a simple color checker tool to ensure that your colors are accurate. The procedure is quite simple, basically you use a standard reference to profile your camera and then your photo editing software makes adjustments to account for any variation.

But it seems to me that camera companies must have already heard about this technology. So it leads me to one of the following thoughts:

a) camera companies do not have the technology to build an ICC profile for their particular sensor and only X-rite (and Datacolor) have this technology. I find this unlikely.

b) camera companies do understand the need for calibrating their cameras, but because of variation from camera to camera, it has to be done individually for each unit, and that is not cost-effective for the camera company, so they leave it to you to do it if its important. That would suggest that there is visible difference between one Nikon D5 and another Nikon D5

c) the color variance is so small that its not worth doing, and that color profiling is only necessary for very few people (but we can all clearly see the variation with our own eyes, so again I think this unlikely)

d) they actually do not wish to have accurate color, for whatever artistic/marketing reasons, and if you want accurate color, you will have to do it yourself. This would also imply that for the majority of photographers, accurate color may not be the ideal choice.

Or is there something else that I'm missing? And yes I realize there is a lot of oversimplification here, like I realize that RAW files have no color its up to the RAW processor to interpret the color, but there must be some variance even then because otherwise RAW processors would simply create a profile for each camera (I mean they do it already) and there would be no need for calibration at all.

Click to expand...

I think they are looking for "pleasing" colour rather then the most accurate colour. Bear in mind that under the camera bonnet, they do the equivalent of external colour profiling, simply to convert the raw data to colour for the jpegs, This process always involves compromises as perfect colour calibration is not possible with a typical RGB camera. I doubt that they would calibrate individual cameras but I suspect the variation in colour response of the Bayer filters from one sensor to another is not that significant.

Dave

MinAZ said:

d) they actually do not wish to have accurate color, for whatever artistic/marketing reasons, and if you want accurate color, you will have to do it yourself.

Click to expand...

Camera manufacturers and software makers do occasionally provide accurate color profiles, ones that match the ColorChecker target as well as they are able.

Photos using these, oddly enough, sometimes appear to be dull and tepid. For example, the version on the left is more accurate than the version on the right:


St. Joachim Church, in Old Mines, Missouri, USA

OK, so folks say that camera makers go for "more pleasing" colors, perhaps like the version on the right, rather than the "more accurate" colors on the left.

But what's often left unsaid is what "more pleasing" actually means. Of course, you can adjust colors anyway way you like as long as it pleases you, but I don't think that is what they mean by that. Instead, there must be a systematic approach to pleasingness which isn't well known or understood, at least by many photographers.

MinAZ said:

Or is there something else that I'm missing?

Click to expand...

Color depends on objective factors, such as the distribution of wavelengths in the light source and the reflection, transmission, and absorption of wavelengths of light by objects in a scene.

Then there are subjective factors, which depend on the observer. If we limit ourselves to only human observers, which is pretty much all we ever care about—maybe with the exception of hunters, who wear blaze orange clothing which is indistinguishable by deer—there still is variation. Color blindness comes in several, well-characterized varieties, and there are changes in color found with those who have had lens replacement due to cataract surgery. There are also transient, psychological color phenomena found in a small percent of the population. But for the most part, a simple, uniform trichromatic color model is assumed for the majority of human beings, and this model is implemented in our cameras, computers and printers. The fact that it works most of the time without much complaint is the best proof for this model.

However, there are also relative factors, which vary by situation even with a single subject. Viewing a scene captured in a tiny photograph gives us a rather different visual experience than viewing the same scene in real life. Suppose you take a photograph of a brightly lit sunlit scene, and then you make a print a view it indoors under artificial lighting: "whites" in the original scene may very well be hundreds or even a thousand times brighter than the same whites in the print: it makes sense in retrospect that this ought to make a difference in perception. Human eye f/stops don't vary all that much, and our visual shutter speed doesn't seem to vary at all, and so we must be left with an intense ISO boost (with heavy noise reduction) which certainly ought to mess with colors, especially considering how little dynamic range will be left.

Understanding how human vision varies by relative circumstance falls under the general name of "Color appearance models", and typical photography only covers the very simplest of models.

For example the Hunt Effect is a visual phenomenon where color saturation appears to decrease with decreasing brightness, and the Stevens Effect likewise causes the appearance of decrease in contrast with decreasing brightness. So it turns out that a photograph *needs* to have boosted saturation and contrast for it to *look* accurate, unless it is displayed in full sunlight.


--

bobn2 said:

MinAZ said:

A thought occurred to me after watching Hudson Henry's camera calibration tutorial. In it, he shows how to use a simple color checker tool to ensure that your colors are accurate. The procedure is quite simple, basically you use a standard reference to profile your camera and then your photo editing software makes adjustments to account for any variation.

But it seems to me that camera companies must have already heard about this technology. So it leads me to one of the following thoughts:

a) camera companies do not have the technology to build an ICC profile for their particular sensor and only X-rite (and Datacolor) have this technology. I find this unlikely.

b) camera companies do understand the need for calibrating their cameras, but because of variation from camera to camera, it has to be done individually for each unit, and that is not cost-effective for the camera company, so they leave it to you to do it if its important. That would suggest that there is visible difference between one Nikon D5 and another Nikon D5

c) the color variance is so small that its not worth doing, and that color profiling is only necessary for very few people (but we can all clearly see the variation with our own eyes, so again I think this unlikely)

d) they actually do not wish to have accurate color, for whatever artistic/marketing reasons, and if you want accurate color, you will have to do it yourself. This would also imply that for the majority of photographers, accurate color may not be the ideal choice.

Or is there something else that I'm missing? And yes I realize there is a lot of oversimplification here, like I realize that RAW files have no color its up to the RAW processor to interpret the color, but there must be some variance even then because otherwise RAW processors would simply create a profile for each camera (I mean they do it already) and there would be no need for calibration at all.

Click to expand...

I think the answer is hanging somewhere around (d). The market doesn't want accurate colour, it wants accentuated colour. Even in the days of colour film this was true, and films which offered very true to life colour tended not to last long.

Click to expand...

Accurate colour is needed for photographing works of art, products, textile designs, etc. It is not, usually, what is wanted in general photography.

TacticDesigns said:

Klaus dk said:

...

I often wonder how colour can survive at all in the chain starting with crappy artificial light over subjects painted or made up with artificial dyes, through three-colour Bayer filters, algorithmic conversion, jpeg viewers, uncalibrated screens of uncertain characteristics to our eyes where the whole picture is once again filtered into a combination of three colour channels before it is processed into a perception.

Click to expand...

+1

Or printed on printers that have their own characteristics.

This is something that I have to deal with with paid jobs.

It is expected that I produce pleasing pictures.

For that I use X-Rite equipment.

When I shoot portraits, I'll use an X-Rite Colorchecker Passport as a target.

And I have my monitor calibrated with an X-Rite Color Munki Display.

And I have the printer profile from the commercial printer I submit pictures to installed in Lightroom so I can do a soft proof to simulate what this image would look like printed on their printer.

So . . .

1. I take a picture of the X-Rite Colorchecker Passport while doing the portait session.

2. I load that picture up in Lightroom. Export as DNG. Use the X-Rite software to create a profile. Load that back in to LR. And then apply that profile to the shots from that session.

3. I then edit the pictures in LR on my colour calibrated monitor, that I calibrated with X-Rite Color Munki Display.

4. And then I preview what it would look like printed with the printer profile I received from the commercial printer I use, that has been loaded into my copy of LR.

I may make tweaks based on personal preference.

But, using this equipment gets me to the starting line really quickly!

Take care & Happy Shooting!

Click to expand...

Do the customers like the results ?

My impression is that many people prefer skin colours that are not quite accurate.

petrochemist said:

I think the idea of accurate color is flawed. To accurately access the color you need a spectrometer.

Unless the spectral response of the red,green & blue channels of the output device exactly match those of the camera it is impossible to get accurate color for all viewers.

The human eye is not a standard device which works the same for everyone. Some people have some sort of color blindness so that specific colors are indistinguishable, then you have variations in how far into the UV & IR ends of the spectrum people can see...

Cameras are set to be able to give reasonably accurate colors to the average human, they do quite well at this, but wouldn't stand a chance of fooling other creatures which often see more colors than we do - A mantis shrimp apparently has 13 types of color receptor & IIRC a horse has 5...

Click to expand...

I think you are remembering wrongly about horses. Like most mammals, they have only two types of cone, and by human standards they are "colour blind" dichromats.