Nikon D80 colorspace surprise

[tom_halfhill]

Recently I took some pictures with a Nikon D80 in the default colorspace (Mode

I, sRGB) and compared them with pictures of the same subjects taken in one of

the custom colorspaces (Mode II, Adobe RGB). Supposedly, Adobe RGB has a wider

color gamut than sRGB and should produce more accurate color. I wasn't expecting

much difference.

 

To my surprise, almost all my comparison photos showed a noticeable difference,

and some showed a dramatic difference. These differences became much greater

when I printed the images at Walgreens. The differences were so great that I

returned to the original scenes in the same light and compared them with the

prints. In every case, the sRGB colors were more accurate than the Adobe RGB colors.

 

All pictures were taken with automatic white balance in bright mid-day sun,

center-weighted metering, aperture-priority mode. The exposures were perfect.

Auto-adjusting the levels in Photoshop had almost no visible effect on either

the sRGB or Adobe RGB images.

 

I'm posting one example comparison below. The color difference is clearly

visible on my (noncalibrated) screen, and it's much greater in the Walgreens

prints. I've heard that the Fuji Frontier printers at Walgreens are calibrated

to favor sRGB, but the color difference is visible on screen as well, and sRGB

beats Adobe RGB every time.

 

Can someone explain why a larger color gamut produces less accurate color?

(Admittedly, my experiment was limited; I didn't shoot pictures in anything but

bright sunlight.)<div></div>

[ellis_vener_photography]

"Supposedly, Adobe RGB has a wider color gamut than sRGB and should produce more

accurate color. I wasn't expecting much difference."

 

Adobe RGB(1998) is larger, csaturated colors aren't clipped (as much) . That doesn't

translate on this or any other planet as "more accurate".

 

"To my surprise, almost all my comparison photos showed a noticeable difference, and

some showed a dramatic difference. These differences became much greater when I

printed the images at Walgreens."

 

Their printers are probably dumbed down to handle sRGB over what they are really capable

of. A printer profile is very different from a device independent workspace (sRGB, Adobe

RGB(1998), Pro Photo, etc.). And also there are more than a few people who don't believe

that what Nikon codes as sRGB and calls Adobe RGB(1998) actually follow the CIE specs.

 

Is your monitor accurately calibrated and profiled?

 

Mine is and the sRGB example looks way over saturated in the blues and greens.

 

But hey, you don't work on my computer and I don't take your photographs so if what

works for you works for you but don't mistake that

[Hamor Photography]

<p>Welcome to the wonderful world of color profiles. Please sit down, relax, and gouge

your brain out with a spoon. It's the only way to keep from going crazy.</p>

 

<p>But seriously, your results appear normal to me from my experience shooting RAW vs.

JPEG Adobe RGB vs. JPEG sRGB on my Canon gear.</p>

 

<p>Wider gamut just means there's theoretically more data to play with.</p>

 

<p>Read Ken Rockwell's <a href="http://www.kenrockwell.com/tech/adobe-

rgb.htm">sRGB vs. Adobe RGB</a> and then follow that up with <a href="http://

www.kenrockwell.com/tech/color-management/is-for-wimps.htm">Color Management is

for Wimps</a>.</p>

 

<p>Also, your printing experience is on par; virtually all consumer printing services

expect sRGB and will fudge other profiles.</p>

[Tim_Lookingbill]

First off it needs to be established whether you are viewing these

images in a color managed application and if so, what display

profile is loaded in the system? It needs to be accurate.

 

Now, assuming both images look different in let's say

Photoshop it's still understandable that you'ld get two different

renderings of the same image.

 

Several years ago over at the Adobe Mac Photoshop forums a

professional photographer once posted on the same issue with

a high end Canon (can't recall the model). But its color rendering

of two identical exposures of a GretagMacbeth color chart taken

one after the other showed the opposite-higher saturation in the

AdobeRGB version.

 

No one could explain what was going on, so I established a

hypothesis that maybe the rendering algorithms within the

camera's processes weren't being cleared within the camera's

cache when both shots were taken one after the other.

Subsequently this algorithm was applied to both images

mapping identical sensor data to two different color spaces

much the same as applying identical edits in Photoshop to two

identical images with different color spaces.

 

Try it to see what I mean. Make a duplicate of one image already

in one color space and convert it to the other and apply color and

luminance changes, save it and apply it to the other image. You'll

get two different renderings.

 

Color spaces will only affect color mapping algorithms as they

relate to your display profile in CM apps. The bigger color space

only gives you more headroom in regards to clipping RGB

numbers.

 

Another guess as to what's happening within the camera is

Nikon's algorithms may have been hardwired within the camera

to intentionally map color differently between the two spaces as

suggested by Ellis.

[tom_halfhill]

Ellis: "Adobe RGB(1998) is larger, csaturated colors aren't clipped (as much). That doesn't translate on this or any other planet as 'more accurate.'"

 

Why shouldn't less clipping be more accurate? Clipping discards data. More color data should produce more accurate color. Otherwise, why do we need wider color gamuts?

 

Ellis: "[Walgreens] printers are probably dumbed down to handle sRGB..."

 

Yes, I noted that.

 

Ellis: "Is your monitor accurately calibrated and profiled?"

 

As my original message said, no it isn't.

 

Ellis: "Mine is and the sRGB example looks way over saturated in the blues and greens."

 

Yep, the sRGB example looks oversaturated on my uncalibrated monitor, too. But as I said in my original message -- which you seem to have skimmed rather quickly -- the "oversaturated" photo is obviously more accurate when I return to the original scene and compare it with the print. Maybe we have brighter colors in California than you are used to seeing.

 

Ellis: "But hey, you don't work on my computer and I don't take your photographs so if what works for you works for you..."

 

I'm not asking for a value judgment of which photo has the colors you or I prefer. I'm asking why the smaller color gamut produced more accurate color. I am able to compare the photos with the actual scene. All you have is a calibrated monitor. There is no doubt that the "oversaturated" sRGB photo is a more faithful rendering of the scene. Why is that so?

[Hamor Photography]

<blockquote><em>There is no doubt that the "oversaturated" sRGB photo is a more

faithful

rendering of the scene. Why is that so?</em></blockquote>

 

<p>Did you read my above links to Ken Rockwell's articles? This sentence basically

summarizes both articles:</p>

 

<blockquote><em>Adobe RGB squeezes colors into a smaller range (makes them duller)

before recording them to your file.</em></blockquote>

 

<p>So Adobe RGB is better <em>in theory</em> if you're immediately going into

Photoshop for retouching and color correction <em>before</em> exporting to sRGB for

the Web or printing 4x6 snapshots. But if you're planning on using the majority of your

shots straight from the camera then just stick with sRGB. I've found that I can use most

sRGB shots right from the camera but have to color correct almost anything I've shot in

Adobe RGB.</p>

[bruce_rubenstein___nyc]

What's going on here with the D80 (and D200 for that matter) has NOTHING to do with the inherent differences between sRGB (Nikon's Ia) and AdobeRGB (II). It has to do with how Nikon renders the colors. Nikon increases the saturation of red and shifts the hue of blue in Ia to give "better" looking color. Here's a link to detailed info: http://www.imaging-resource.com/PRODS/D80/D80IMATEST.HTM

 

It also explains why I thought the skin tones from my D200 & D80 looked too ruddy when shot in sRBG, JPGs. I shot a Macbeth color chart in JPG, in the two color spaces and the AdobeRGB is more accurate. But as Nikon, and other camera makers know, just like the film companies know, not everyone realy wants accurate.

[Tim_Lookingbill]

Color spaces were originally invented to accomodate the setting

up of automated color mapping processes between devices for

commercial press output. Where the big money's at. It doesn't

mean they have more color, they just map color with varying

characteristics according to how they will map to a device like a

monitor or others with much less colors to work with like from a

commercial press. You come up with a million colors using only

four inks on absorbant paper.

 

Photoshop is a toolbox for ALL digital imaging processes, not

just for photographers who most don't care squat about

commercial press output. Color spaces allow a lining up of color

mapping processes like a bowsman lining up his arrow to nail

color targets like offset 100% yellow and cyan or Ektaspace

oranges, blues, teals and fleshtones with the least amount of

image correcting and headaches.

 

If you want to understand the differences/advantages of color

spaces just come up with tints of blues and oranges and 100%

CMYK yellow in AdobeRGB and sRGB using the color picker.

You'll soon start seeing differences in the appearance between

the two spaces as evidenced in the color transitions within the

color window.

 

To even see it more pronounced try comparing EktaspaceRGB

to sRGB and come up with a decent cadmium yellow without

clipping RGB data but still allow it to map SWOP 100% yellow.

 

The entire imaging industry is trying to get everyone on the same

page especially photographers as to the meaning of color. Color

spaces act only as a tool to getting there.

 

If you like what you get using sRGB out of your camera, use it,

but it's a crappy editing space. But your incamera's processing

must now always be a talented marksman when it comes to

scene rendering for all the types of scenes you'll be capturing.

[stefan_g]

You are not saying whether you 'tell' the printer that the Adobe RGB image is in the Adobe RGB color space. IF you are keeping the image originally saved in-camera as Adobe RGB in that space (i.e. you are not saving to a different color space in e.g. Photoshop) and then print it without embedded profile (or with the profile being ignored) on a printer that assumes sRGB unless told otherwise, then the printer will tone everything down. The same real world color will be encoded using smaller numbers in the larger (Adobe RGB) space that allows for a wider gamut, and then gets misinterpreted in the smaller (sRGB) space as meaning a less saturated real world color.

 

I think this (interpreting Adobe RGB numbers as sRGB) should dominate the color shift; if I understand Bruce's post correctly, there is also the secondary effect of how the mapping from real world color to camera space might be tweaked differently for different spaces by the camera manufacturer.

 

The same 'misinterpretation argument' applies for viewing on screen in a non color-managed application; if you are viewing in a color-managed application (and are using it correctly), then only the mechanism suggested by Bruce comes into play.

[tom_halfhill]

After reading the explanations posted here, and Ken Rockwell's articles, I'm still a little confused. (Sorry.) Some people have explained that Adobe RGB colors appear less saturated when the file is converted to the sRGB colorspace. But where is that conversion happening? It's not just a Walgreens effect, because the Adobe RGB colors look less saturated on my screen, before printing. Is Photoshop automatically converting Adobe RGB files to sRGB?

 

Maybe if I explain my test procedure in more detail, someone can straighten it out.

 

1. I shot two pictures of each scene -- one with the camera set for color mode I (sRGB, the default colorspace), and the other with the camera set for color mode II (Adobe RGB). Both in JPEG format.

 

2. I did not print the pictures directly from the camera or memory card. Instead, I copied the photos from the memory card into my computer, then opened them in Photoshop (version 5.0 LE). It might help if someone knows which colorspace this version of Photoshop defaults to. (I assume Adobe RGB.)

 

3. In Photoshop, the two photos look obviously different from each other, before any manipulations. The sRGB photo is significantly more saturated than the Adobe RGB photo.

 

4. I used Photoshop's auto-levels command on both photos, which had a very small effect. In both cases, the shadows got slightly darker (barely noticeable), but the overall exposure and colors didn't change.

 

5. I saved new JPEG files of both photos and uploaded them to the Walgreens website for printing. I didn't explicitly do anything to change the colorspace in Photoshop. Whether Photoshop does that automatically when saving JPEGs, I don't know.

 

6. The color difference between the Walgreens prints (sRGB vs. Adobe RGB) are more pronounced than the screen images are. As I mentioned originally, this is probably because the Fuji Frontier printer at Walgreens prefers sRGB, but it might also be caused by some autocorrection at their end that I can't control. However, the color difference is quite visible on my uncalibrated monitor, and the Walgreens prints are very close to my screen images, so I think my monitor is pretty close to proper calibration. (I have printed hundreds of other images at Walgreens with 99% good results.)

 

7. Surprised by the color difference, I returned to the actual scene and compared it with the prints. The sRGB version is definitely more accurate -- a more faithful rendering of the actual colors.

 

So we have some unknown variables:

 

- The camera's color mapping to different colorspaces

 

- The camera's internal RAW-to-JPEG conversion

 

- Photoshop LE 5.0's default colorspace when opening JPEGs

 

- Photoshop LE 5.0's default colorspace when saving JPEGs

 

- Walgreen's Fuji Frontier default colorspace (probably sRGB)

 

- Walgreen's Fuji Frontier autocorrection settings

 

But again, the last two variables don't explain the visible difference between the images on screen, in Photoshop.

[Tim_Lookingbill]

At first opening in Photoshop 5 LE are you retaining the file's

color spaces and not stripping them of their profiles when the

dialog box gives you those options?

 

Check your RGB Working Space in Color Settings under the Edit

menu and make sure Color Management Policies for RGB is set

to Preserve Embedded Profiles and all the little boxes

underneath are checked.

 

Your first two unknown variables seem to be the answer here.

The majority of Frontier minilabs and minilabs in general print to

an sRGB space except for certain colors that near 0 and

255RGB in any color combination.

[gerund]

The unknown variable that you are looking for may well be your monitor.

Unless you are using a high grade monitor such as an Eizo that is capable of resolving the adobe rgb colorspace and have the monitor adjusted for that colorspace, you will not be able to soft proof that colorspace successfully in photoshop.

[oharritz]

I've been at it for 6 years, that is digital photography, and I've developed a pretty good understanding of color profiling.

I'am printing fine art prints with an color profiled Epson but only using sRGB.

3 years ago i decided that adobeRGB was not worth the effort, since the result in a finished print was minimal. This may ofcourse change when the inkjet technology gets better.

 

So the short answer is... If you have an expensive corretly profiled CRT monitor, not older than 3 years, and an inkjet that can handle the extended gamut, and a lot of time to get it right, then you are ready to persue the wonders of adobeRGB or better color space.

If not, forget it and use your energy and time to get results in sRGB which in most cases are just as good as adobeRGB and in rare cases almost as good.

Regards, and happy shooting.

[tom_halfhill]

Tim said: "At first opening in Photoshop 5 LE are you retaining the file's color spaces and not stripping them of their profiles when the dialog box gives you those options?"

 

No such dialog box appears.

 

Tim said: "Check your RGB Working Space in Color Settings under the Edit menu and make sure Color Management Policies for RGB is set to Preserve Embedded Profiles and all the little boxes underneath are checked."

 

There are no such options under the Edit menu or anywhere else that I can find.

 

And that probably solves the mystery. I'm guessing that Photoshop LE 5.0 doesn't support Adobe RGB. It was probably one of the features omitted from LE (Limited Edition). LE likely defaults to sRGB, or whatever the system (in this case, Windows XP) is using.

 

It doesn't matter much to me...I am getting accurate color with sRGB. And because my system is uncalibrated, and I'm sending output to either a consumer-level inkjet printer or the Fuji Frontier at Walgreens, then sRGB is the better choice.

 

After all, the point of my little experiment was to see whether sRGB or Adobe RGB delivered better results for me. I was surprised by the results, which is why I posted here.

 

Thanks for your help.

[Hamor Photography]

<p>I think I found your answer. :)</p>

 

<p><a href="http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=0018HU">http://

photo.net/bboard/q-and-a-fetch-msg?msg_id=0018HU</a></p>

 

<p><blockquote><em>I went to the Adobe forums and found out that Photoshop 5 LE does

not have

colour management and that the gamma is hard coded to 1.8.</em></blockquote></p>

[tom_halfhill]

Thanks, Sean. That is what I concluded, too. I think it's odd that a version of Adobe Photoshop -- even the "limited edition" version -- does not support Adobe RGB. I wonder if Adobe Elements does? Probably not.

[brucecyr]

Photoshop Elements supports Adobe RGB.

[User_503771]

Industry standardization of color reckoning is fairly recent.

 

So, you're not getting an accurate rendering of Adobe RGB in your version of Photoshop.

 

I've recently been working with this stuff myself. I upgraded straight from Photoshop version 5 to CS2. Now I'm working out my calibration problems (an entirely different matter).

 

I suspect that your Adobe RGB print looks much different than the sRGB print because the printing device did not notice the difference in profiles, and thus used the sRGB color numbers when printing from the Adobe RGB file, which would have different numbering for its colors.

 

When I make a photo in the Adobe RGB mode, I find that I get subtleties in the color that I can't get in sRGB. Both are accurate within their limits.

 

If you're familiar with black and white film photography, you might best recognize the problem through an analogy:

 

The human eye can capture (and accurately "display") a much greater range of luminance than film, which can capture a much greater range of luminance than printing paper. Ansel Adams went to a lot of trouble codifying this. Thus, his "profile" for a grade 0 paper (wide range of luminance rendering) was much different than that for a grade 3 or 5 paper (narrower range of luminance). He would expose and develop the negative such that he would be able to capture its luminance range on the printing paper.

 

Various color profiles are optimized for various printing devices. sRGB was developed to enhance the chances of getting acceptable color out of typical consumer printing devices -- both for the home and at photo labs such as the ones at Walgreen and other popular stores. sRGB is ideal for the snapshot market, while being versatile enough to also meet the needs of many serious amateur artists.

 

Adobe RGB and Prophoto profiles are examples which offer expanded color "gamut" (more colors) for higher-end printing (or display) devices which can accurately reproduce a greater range of colors. When printing from one to another, "profile-aware" software will adequately approximate the "out-of-range" colors in the destination profile.

 

So, the standardized profiles work with software which knows the difference between them, and thus can "translate" the colors of one profile into the appropriate profile for the device being used.

 

If the Walgreen output device had known you had presented it with an Adobe RGB file, it would have converted those colors into sRGB at print time, and you would have gotten two nearly-identical prints.

 

A ton of work has gone into all of this, starting with coming to terms with how to define one color, distinct from another color, in a way that everyone concerned could understand and use -- and also agree with!

 

One excellent piece of software is free from Microsoft -- WinColor.exe. I downloaded it from the Microsoft site (I don't recall the link, but I did manage to find it with a search. You might google it....)

 

If you run it, click on the "profiles" tab and you can compare various profiles that are installed on your system. If you have even only a few profiles, that's all you need.

 

If you don't have any, you can download them from various sources -- monitor manufacturers, inkjet paper manufacturers (HP, Ilford, et al). It's a great piece of software for comparing color profiles, and provides a great "demo" of the whole color profile shenanigan.

 

Hope this helps.