Is Nikon Mode III really sRGB?

Started Apr 6, 2005 | Discussions
Bill Janes Senior Member • Posts: 1,848
Is Nikon Mode III really sRGB?

Nikon refers to Modes I and III as sRGB, implicitly recognizing two variants of sRBG, and Mode II as AdobeRGB. How do these RGBs compare to the Adobe Photoshop sRGB IEC 61966-2.1, which I presume is the standard?

http://www.pbase.com/bjanes/image/41639470.jpg/large
--
Bill Janes

Valbo Contributing Member • Posts: 687
Re: Is Nikon Mode III really sRGB?

I think you mix up the color spaces "Adobe RGB 1998" and "sRGB IEC 61966-2.1".

There is not ONE standard, they are both standards.

-- hide signature --

Best regards
Valbo

Bill Janes wrote:

Nikon refers to Modes I and III as sRGB, implicitly recognizing two
variants of sRBG, and Mode II as AdobeRGB. How do these RGBs
compare to the Adobe Photoshop sRGB IEC 61966-2.1, which I presume
is the standard?

http://www.pbase.com/bjanes/image/41639470.jpg/large
--
Bill Janes

OP Bill Janes Senior Member • Posts: 1,848
Re: Is Nikon Mode III really sRGB?

Valbo wrote:

I think you mix up the color spaces "Adobe RGB 1998" and "sRGB IEC
61966-2.1".

There is not ONE standard, they are both standards.

I was referring to Nikon Mode III, which I don't think is a standard. Of course the two spaces you referred to are standards.

My message was posted prematurily while I was editing it. See the complete message which I re-posted.

Bill Janes

-- hide signature --

Best regards
Valbo

Bill Janes wrote:

Nikon refers to Modes I and III as sRGB, implicitly recognizing two
variants of sRBG, and Mode II as AdobeRGB. How do these RGBs
compare to the Adobe Photoshop sRGB IEC 61966-2.1, which I presume
is the standard?

http://www.pbase.com/bjanes/image/41639470.jpg/large
--
Bill Janes

misifus Regular Member • Posts: 137
Re: Is Nikon Mode III really sRGB?

Bill Janes wrote:

Valbo wrote:

I think you mix up the color spaces "Adobe RGB 1998" and "sRGB IEC
61966-2.1".

There is not ONE standard, they are both standards.

I was referring to Nikon Mode III, which I don't think is a
standard. Of course the two spaces you referred to are standards.

Adobe RGB and sRBG are standards for photo color information to be stored in a file. The difference between mode I and mode III, as I understand it, is one of color sensitivity. The two modes represent different sensitivities to the various portions of the spectrum. Having captured the data, it is then stored according to the sRBG protocals. So, both I and III are sRBG.

-Raf

-- hide signature --

CP900
CP5700
D-70

John Yu Senior Member • Posts: 1,201
Re: Is Nikon Mode III really sRGB?

Nikon has made it confusing. The colour modes don't only change the colorspace, but also how the in-camera image processor interpret the colours in the raw data.

Both mode I and mode III produces the data in the same sRGB colorspace. But the output colours are generated differently. for example, a particular shade of red is rendered as sRGB(240, 2, 2) in mode I, it is rendered as sRGB(231, 1, 1) in mode III. (I made up the values in this example. They're not real.)
--
John

OP Bill Janes Senior Member • Posts: 1,848
Re: Is Nikon Mode III really sRGB?

John Yu wrote:

Nikon has made it confusing. The colour modes don't only change the
colorspace, but also how the in-camera image processor interpret
the colours in the raw data.

Both mode I and mode III produces the data in the same sRGB
colorspace. But the output colours are generated differently. for
example, a particular shade of red is rendered as sRGB(240, 2, 2)
in mode I, it is rendered as sRGB(231, 1, 1) in mode III. (I made
up the values in this example. They're not real.)
--
John

John,

I think that your explanation is correct, but Nikon appears to be using non-standard color spaces. If we are shooting in raw the actual data in the stored file is the same regardless of the mode, since the files are merely tagged with the selected mode.

As I understand things, the generation of the output values of RGB is done according to the selected color space. The actual math can be a little convoluted, but the definition of the space is pretty simple. It involves specifying the gamma, white point, and primaries in CIExyY as shown in this screen capture from PhotoshopCS.

Interestingly, the olny difference between sRGB and aRGB is in the green primary, where aRGB has extended range. Since mode III shows green to advantage, it must be using a diferent green primary than is used in Mode I, more like aRGB.

These details are explained in great detail on Bruce Lindbloom's site, and he shows how to convert between spaces and has an online calculator to do the math.

http://www.brucelindbloom.com/

My assumption that Adobe's sRGB was standard was incorrect. They use a simplified sRGB for gamma correction. In the actual sRGB specification a linear equation is used for low values, and a power function with a gamma of 2.5 is used for higher values to give an average gamma of 2.2. Adobe sRGB and aRGB use a gamma of 2.2 for all values. I don't know what Nikon does, but it appears to be nonstandard.
--
Bill Janes

(unknown member) Forum Pro • Posts: 13,189
Why multiple sRGBs

Funny, someone just emailed me off list about this so I'm going to be lazy and cut and paste:

No camera initially produces sRGB. The RAW data is a Grayscale file. When you ask the camera to provide a file in sRGB there are two processes going on. The first is called rendering. The Grayscale RAW to color image (known as demosaicing) happens. The rendering it totally up to the manufacture to decide how they feel they are producing the most pleasing color. If you shot an identical scene with a Nikon and a Canon in sRGB, more than likely the rendering would not be the same (maybe not even close). This is much like the perceptual intent in ICC printer profiles. You can measure the same target and build two different profiles and get two different results. The manufacture has control over this rendering. Note that when you bring a RAW file into a RAW converter, this rendering process is now under your control. Just as one RAW converter might have better (or worse) default rendering, you the user can control this.

The next step is called encoding. The rendered color is encoded into a color space. This is fixed and non ambiguous. Two identical renderings will have identical encoding into a color space.

You'll note that many DSLRs have several settings for sRGB. This is just a tweak to the rendering phase of the process. Think if it like a "Velvia" setting versus an "Ekatchrome" setting whereby the manufacturers are again not trying to produce a colorimetrically accurate representation of the scene but rather the scene rendered using various bias based on that image appearing on a display. IOW, when you shoot an image of the scene, you're getting the representation of that scene as it would appear on an sRGB display.
-------

OK. I'm back. Note that when I say colorimetrically accurate representation, I'm talking the measured color of the scene (which includes stuff like the Illuminant, the dynamic range and so forth). Colorimetrically correct color looks pretty ugly when viewed on a display. It needs to be rendered (something called output referred) to appear as we hope the scene to appear on this output device.

What's interesting is few reviewers tell you how well the manufacturers are doing with this on the fly rendering. And if they compared the color based on RAW data, you'd have to factor in their skill on rendering using some RAW converter (which one? Each plays a role here).
--
Andrew Rodney
Author of Color Management for Photographers
The Digital Dog
http://www.digitaldog.net

RFoskett Junior Member • Posts: 36
Re: Why multiple sRGBs

My first post so please bear with me. I currently shoot in jpeg but will be moving to raw now that I have PS CS. I post all my images in PS before I have them printed.

Where should I be settig my camera, sRGB or Adobe RGB. I was under the impression that if I did post production in PS that I should shoot in Adobe RGB. With my daughter's wedding coming up I need your direction to make the right decision.

Thanks
Bob

John Yu Senior Member • Posts: 1,201
Re: Is Nikon Mode III really sRGB?

HI Bill,

I'm not a guru in colorspace. But I suspect tagging of an image with a colorspace is more than those few parameters you mentioned. I have the experience of converting and tagging an image with a custom printer profile. The JPG file size grew significantly because of the embedded colorspace metadata. I don't have PS CS around me now. But it looks from your screenshot the custom colorspace you were constructing had to be based on an existing colorspace (See "Primaries" on the screenshot). It looks like the new colorspace will get its parameters and data from that existing one.

Btw, I think you're right that Nikon's sRGB, or even aRGB, are slightly different from those in PS.
--
John

Bill Janes wrote:

John Yu wrote:

Nikon has made it confusing. The colour modes don't only change the
colorspace, but also how the in-camera image processor interpret
the colours in the raw data.

Both mode I and mode III produces the data in the same sRGB
colorspace. But the output colours are generated differently. for
example, a particular shade of red is rendered as sRGB(240, 2, 2)
in mode I, it is rendered as sRGB(231, 1, 1) in mode III. (I made
up the values in this example. They're not real.)
--
John

John,

I think that your explanation is correct, but Nikon appears to be
using non-standard color spaces. If we are shooting in raw the
actual data in the stored file is the same regardless of the mode,
since the files are merely tagged with the selected mode.

As I understand things, the generation of the output values of RGB
is done according to the selected color space. The actual math can
be a little convoluted, but the definition of the space is pretty
simple. It involves specifying the gamma, white point, and
primaries in CIExyY as shown in this screen capture from
PhotoshopCS.

Interestingly, the olny difference between sRGB and aRGB is in the
green primary, where aRGB has extended range. Since mode III shows
green to advantage, it must be using a diferent green primary than
is used in Mode I, more like aRGB.

These details are explained in great detail on Bruce Lindbloom's
site, and he shows how to convert between spaces and has an online
calculator to do the math.

http://www.brucelindbloom.com/

My assumption that Adobe's sRGB was standard was incorrect. They
use a simplified sRGB for gamma correction. In the actual sRGB
specification a linear equation is used for low values, and a power
function with a gamma of 2.5 is used for higher values to give an
average gamma of 2.2. Adobe sRGB and aRGB use a gamma of 2.2 for
all values. I don't know what Nikon does, but it appears to be
nonstandard.
--
Bill Janes

-- hide signature --

John

John Yu Senior Member • Posts: 1,201
Re: Why multiple sRGBs

Hi Bob,

Do you mean you're going to use Adobe Camera Raw (ACR) for the RAW conversion? If so, the colour mode set on the body is redundant and ignored by ACR.You can set the output colorspace on the ACR screen.
--
John

RFoskett wrote:

My first post so please bear with me. I currently shoot in jpeg
but will be moving to raw now that I have PS CS. I post all my
images in PS before I have them printed.

Where should I be settig my camera, sRGB or Adobe RGB. I was under
the impression that if I did post production in PS that I should
shoot in Adobe RGB. With my daughter's wedding coming up I need
your direction to make the right decision.

Thanks
Bob

-- hide signature --

John

OP Bill Janes Senior Member • Posts: 1,848
Re: Is Nikon Mode III really sRGB?

John Yu wrote:

HI Bill,

I'm not a guru in colorspace. But I suspect tagging of an image
with a colorspace is more than those few parameters you mentioned.
I have the experience of converting and tagging an image with a
custom printer profile. The JPG file size grew significantly
because of the embedded colorspace metadata. I don't have PS CS
around me now. But it looks from your screenshot the custom
colorspace you were constructing had to be based on an existing
colorspace (See "Primaries" on the screenshot). It looks like the
new colorspace will get its parameters and data from that existing
one.

John,

I'm not a color space guru either and had to do a little research to respond to your posting and learned a few things in the process. Thank you for your input.

You must differentiate color space definition profiles from device profiles. The color space profiles are defined by the parameters shown in the Photoshop screen shots that I published, and can be converted mathematically by applying a linear 3*3 matrix transformation as shown on Bruce Lindholm's site. Look under the math tab.

http://www.brucelindbloom.com/

I did a search of the icm and icc files on my drive and sorted them by size.

The Adobe RBG icm is 540 bytes and the ProPhotoRGB icm is 940 bytes so the file size is not significantly afftected by whether or not you save the profile with the file.

On the other hand, the Drycreek icc profile for my local Costco Noritsu printer is about 1,400,000 bytes. Printers are not well behaved linear devices and the transform for these devices is usually done by means of a color lookup table (CLUT) which lists the color values for each input and output value. The transform is done by looking up the corresponding output value for the input value. Since a complete CLUT could become quite large (especially with high bit files), usually representative values are listed and some type of interpolation is done. The icm profiles for my Epson 2200 printer are about 110,000 bytes, apparently less comprehensive than the Drycreek profiles.

I found a number of Nikon icm files in the range of about 220,000 bytes. I don't know what these are used for, nor do I know the difference between icm and icc files. Perhaps Nikon Capture uses these in defining modes Ia and IIIa.

Btw, I think you're right that Nikon's sRGB, or even aRGB, are
slightly different from those in PS.
--
John

Bill

OP Bill Janes Senior Member • Posts: 1,848
Re: Why not multiple aRGBs?

Andrew,

Your post cleared up a lot of my misunderstanding of the profusion of sRGB settings on digital cameras. The renderings are different, but they all encode to the standard sRGB space.

Most experts (e.g. Bruce Fraser) recommend shooting in the raw mode and rendering into a wider color space such as Adobe RGB or ProPhotoRGB when the images are to be used for printing, since the sRGB space can't represent the gamut of a good digital camera.

Nikon cameras that I am familiar with have only one setting for aRGB. You can adjust contrast and saturation in Nikon Capture or Adobe Camera Raw, but selective enhachment of specific colors such as done in Mode IIIa of Nikon is not so easily done. Should cameras have selections for rendering into the aRBG space?

I see that you have a new book on color management for photographers. I had been intending to get Bruce Fraser's Real World Color Management book, but now I'll have to take a look at your book too.

Best regards,

Bill Janes

digidog wrote:

Funny, someone just emailed me off list about this so I'm going to
be lazy and cut and paste:

No camera initially produces sRGB. The RAW data is a Grayscale
file. When you ask the camera to provide a file in sRGB there are
two processes going on. The first is called rendering. The
Grayscale RAW to color image (known as demosaicing) happens. The
rendering it totally up to the manufacture to decide how they feel
they are producing the most pleasing color. If you shot an
identical scene with a Nikon and a Canon in sRGB, more than likely
the rendering would not be the same (maybe not even close). This is
much like the perceptual intent in ICC printer profiles. You can
measure the same target and build two different profiles and get
two different results. The manufacture has control over this
rendering. Note that when you bring a RAW file into a RAW
converter, this rendering process is now under your control. Just
as one RAW converter might have better (or worse) default
rendering, you the user can control this.

The next step is called encoding. The rendered color is encoded
into a color space. This is fixed and non ambiguous. Two identical
renderings will have identical encoding into a color space.

You'll note that many DSLRs have several settings for sRGB. This is
just a tweak to the rendering phase of the process. Think if it
like a "Velvia" setting versus an "Ekatchrome" setting whereby the
manufacturers are again not trying to produce a colorimetrically
accurate representation of the scene but rather the scene rendered
using various bias based on that image appearing on a display. IOW,
when you shoot an image of the scene, you're getting the
representation of that scene as it would appear on an sRGB display.
-------

OK. I'm back. Note that when I say colorimetrically accurate
representation, I'm talking the measured color of the scene (which
includes stuff like the Illuminant, the dynamic range and so
forth). Colorimetrically correct color looks pretty ugly when
viewed on a display. It needs to be rendered (something called
output referred) to appear as we hope the scene to appear on this
output device.

What's interesting is few reviewers tell you how well the
manufacturers are doing with this on the fly rendering. And if they
compared the color based on RAW data, you'd have to factor in their
skill on rendering using some RAW converter (which one? Each plays
a role here).
--
Andrew Rodney
Author of Color Management for Photographers
The Digital Dog
http://www.digitaldog.net

(unknown member) Forum Pro • Posts: 13,189
Re: Why not multiple aRGBs?

Bill Janes wrote:

Nikon cameras that I am familiar with have only one setting for
aRGB. You can adjust contrast and saturation in Nikon Capture or
Adobe Camera Raw, but selective enhachment of specific colors such
as done in Mode IIIa of Nikon is not so easily done. Should cameras
have selections for rendering into the aRBG space?

They could in theory but I don't know any that do. Well wait, the Olympus E-10 has separate settings that apply to all the color spaces rendered from the RAW data on the fly. There's no reason this can't happen. I suspect it's not done to keep thinks simpler for the user. The companies might assume if someone is using Adobe RGB (1998), they are savvy enough to shoot RAW and do whatever they wish.

If we could time travel a few years into the future when the vast majority of displays will be able to produce the Adobe RGB (1998) gamut and behavior as we now have with sRGB, I suspect sRGB will be used less and less. We'll see what happens on the various cameras by then.
--
Andrew Rodney
Author of Color Management for Photographers
The Digital Dog
http://www.digitaldog.net

OP Bill Janes Senior Member • Posts: 1,848
Re: Why not multiple aRGBs?

Andrew,

After reading your post I spent some time at the Drycreek interactive gamut comparison site. For those not familiar with this site, it shows 3D wire frame models where you can select the editing space, camera gamut, and output space (either monitor or printer).

The gamut of even an entry level digital SLR such as the Nikon D70 is surprisingly wide, and the ProPhotoRGB is required to capture the full gamut of the camera.

They list a few high end monitors which can show most of the aRGB gamut as you mentioned.

What was surprising to me was the limited gamut of the contone photo printers (Fuji Frontier, Noritsus, and even the Lightjet). They strain to print even sRGB. What good is to to capture colors that you can't see or print? I guess you have some control of the out of gamut colors with rendering intent.

What surprised me was the gamut of the Epson printers with Ultrachrome inks and the best papers such as Pictorico Hi Gloss White Film. Hopefully you cover some of these issues in your forthcoming book.

Best regards,

Bill Janes

digidog wrote:

Bill Janes wrote:

Nikon cameras that I am familiar with have only one setting for
aRGB. You can adjust contrast and saturation in Nikon Capture or
Adobe Camera Raw, but selective enhachment of specific colors such
as done in Mode IIIa of Nikon is not so easily done. Should cameras
have selections for rendering into the aRBG space?

They could in theory but I don't know any that do. Well wait, the
Olympus E-10 has separate settings that apply to all the color
spaces rendered from the RAW data on the fly. There's no reason
this can't happen. I suspect it's not done to keep thinks simpler
for the user. The companies might assume if someone is using Adobe
RGB (1998), they are savvy enough to shoot RAW and do whatever they
wish.

If we could time travel a few years into the future when the vast
majority of displays will be able to produce the Adobe RGB (1998)
gamut and behavior as we now have with sRGB, I suspect sRGB will be
used less and less. We'll see what happens on the various cameras
by then.
--
Andrew Rodney
Author of Color Management for Photographers
The Digital Dog
http://www.digitaldog.net

(unknown member) Forum Pro • Posts: 13,189
Re: Why not multiple aRGBs?

Bill Janes wrote:

The gamut of even an entry level digital SLR such as the Nikon D70
is surprisingly wide, and the ProPhotoRGB is required to capture
the full gamut of the camera.

Well ya, kind of. I agree the gamuts can be very large. Keep in mind however that digital cameras really don't have a gamut per say (the correct term would be a color mixing function). The profiles you see that describe the gamut of a camera come about when someone photographs a special target and uses that to build a profile. Well the target has it's own gamut. One reason building digital camera profiles is so difficult is we are trying (and usually not well) to define the entire visible world (the scene) using a target. As such, there are lots of colors and tone well outside that target. So while I'm the first to agree that what you're seeing, the size of the camera gamut is large, I suspect it's much larger then these plots represent.

We can photograph a scene with a 10,000:1 contrast ratio (good luck getting more than a 500:1 contrast ratio on today's displays).

If you look at a scanner, it too has a similar color mixing function. But the object we scan (film) does have a specific gamut so it's a lot easier to profile and define them this way.

What was surprising to me was the limited gamut of the contone
photo printers (Fuji Frontier, Noritsus, and even the Lightjet).
They strain to print even sRGB. What good is to to capture colors
that you can't see or print? I guess you have some control of the
out of gamut colors with rendering intent.

What you have to look at is the entire 3D gamut. While yes, there are a great deal of these colors that fall into sRGB, there are colors that don't and this is what we need to be concerned with (assuming we want to print those colors). Even if only 10% of the colors fall outside of the sRGB gamut, are those 10% important to you to express the image you are trying to reproduce?

And while we capture images with potentially huge gamuts today, we have no idea of the gamut of tomorrows output devices. As such, it's probably not a good idea for some to toss away the colors they can capture.

-- hide signature --

Andrew Rodney
Author of Color Management for Photographers
The Digital Dog
http://www.digitaldog.net

RFoskett Junior Member • Posts: 36
Re: Why multiple sRGBs

Thanks very much. I appreciate all the information I have received from this forum.

Bob

Keyboard shortcuts:
FForum MMy threads