Ad
Ad
Ad
Pages: [1] 2 3 ... 6 »   Bottom of Page
Print
Author Topic: Colour spaces  (Read 49748 times)
bobrobert
Guest
« on: November 20, 2007, 09:31:42 AM »
ReplyReply

If someone was to change the colour space of their image from AdobeRGB 98 to pro photo would the colour noise be more apparent and would this necessitate a stronger noise reduction in the chroma settings In a nut shell does a wider colour space mean a more noise in an image?TIA
Logged
bobrobert
Guest
« Reply #1 on: November 21, 2007, 08:26:34 AM »
ReplyReply

Quote
If someone was to change the colour space of their image from AdobeRGB 98 to pro photo would the colour noise be more apparent and would this necessitate a stronger noise reduction in the chroma settings In a nut shell does a wider colour space mean a more noise in an image?TIA
[a href=\"index.php?act=findpost&pid=154389\"][{POST_SNAPBACK}][/a]

I will try and explain it better My work flow up until now was to select Adobe RGB 98 in camera raw prior to processing After reading the benefits of using the pro photo setting in camera raw yesterday I noticed that a 1600 iso image looked more noisy than when it was set to Adobe RGB 98 Was I imagining this? IF not then when using neat image I would have to reduce noise more? Thus degrading the image more? Does a wider colour space mean more noise reduction? Or have I got it all wrong?TIA
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9315



WWW
« Reply #2 on: November 21, 2007, 08:56:35 AM »
ReplyReply

Quote
I will try and explain it better My work flow up until now was to select Adobe RGB 98 in camera raw prior to processing After reading the benefits of using the pro photo setting in camera raw yesterday I noticed that a 1600 iso image looked more noisy than when it was set to Adobe RGB 98 Was I imagining this? IF not then when using neat image I would have to reduce noise more? Thus degrading the image more? Does a wider colour space mean more noise reduction? Or have I got it all wrong?TIA
[a href=\"index.php?act=findpost&pid=154676\"][{POST_SNAPBACK}][/a]

Are you shooting Raw?
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bobrobert
Guest
« Reply #3 on: November 22, 2007, 04:50:46 AM »
ReplyReply

Quote
Are you shooting Raw?
[a href=\"index.php?act=findpost&pid=154683\"][{POST_SNAPBACK}][/a]

Yes Basically I am wondering about noise reduction in different colour spaces If a colour space has a bigger gamut does it need greater noise reductionTIA
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9315



WWW
« Reply #4 on: November 22, 2007, 08:23:31 AM »
ReplyReply

Quote
Yes Basically I am wondering about noise reduction in different colour spaces If a colour space has a bigger gamut does it need greater noise reductionTIA
[a href=\"index.php?act=findpost&pid=154914\"][{POST_SNAPBACK}][/a]

Raw has no color space. You encode a rendering into a color space and hopefully, you can control as much noise reduction in the Raw rendering state as possible prior to encoding.

You DO want higher bit depth in wider gamut spaces, that's for sure.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #5 on: November 22, 2007, 04:34:09 PM »
ReplyReply

Quote
If someone was to change the colour space of their image from AdobeRGB 98 to pro photo would the colour noise be more apparent and would this necessitate a stronger noise reduction in the chroma settings In a nut shell does a wider colour space mean a more noise in an image?TIA
[a href=\"index.php?act=findpost&pid=154389\"][{POST_SNAPBACK}][/a]

No. Unless you do something silly like assigning ProPhoto to a sRGB JPEG. But that will cause major saturation overkill.
Logged

bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #6 on: November 22, 2007, 09:16:33 PM »
ReplyReply

Quote
Raw has no color space. You encode a rendering into a color space and hopefully, you can control as much noise reduction in the Raw rendering state as possible prior to encoding.
[{POST_SNAPBACK}][/a]

That statement is debatable. Look at the Adobe [a href=\"http://www.adobe.com/products/dng/pdfs/dng_spec.pdf]DNG Specifiation[/url] on page 47 where mapping from the camera color space to the CIE XYZ space is discussed. The conversion is done with a three by three matrix, just like a conversion from sRGB to ProPhotoRGB is done. If you look at the source code of DCRaw you can review the actual matrix coefficients and conversion code.

Apparently the folks at Adobe think that the camera has a color space.

Bill
Logged
Hermie
Full Member
***
Offline Offline

Posts: 207


« Reply #7 on: November 23, 2007, 05:49:07 AM »
ReplyReply

Quote
Apparently the folks at Adobe think that the camera has a color space.

Bill
[a href=\"index.php?act=findpost&pid=155113\"][{POST_SNAPBACK}][/a]

Isn't this all about semantics?
The raw data as such don't have a color space in the sense that they have a meaning in terms of PCS, they're just values.
The matrix you describe acts as an input profile to map linearized camera values to PCS.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #8 on: November 23, 2007, 09:09:37 AM »
ReplyReply

Quote
Isn't this all about semantics?
The raw data as such don't have a color space in the sense that they have a meaning in terms of PCS, they're just values.
The matrix you describe acts as an input profile to map linearized camera values to PCS.
[{POST_SNAPBACK}][/a]

If the raw data were to have no meaning with respect to the PCS, then obviously it would not be possible to convert them to the PCS. Since they do have a relationship to the PCS, it is possible to convert from the camera space to CIE XYZ by a standard three by three matrix conversion. Here is an excerpt from the source code of DCRAW:

/*
   Thanks to Adobe for providing these excellent CAM -> XYZ matrices!
 */
void CLASS adobe_coeff (char *make, char *model)
static const struct {
    const char *prefix;
    short black, trans[12];
{ "NIKON D200", 0,
   { 8367,-2248,-763,-8758,16447,2422,-1527,1550,8053 } },

As you can see, these values are a 3 by 3 matrix to convert from the Nikon D200 colorspace to CIE XYZ. Once that is accomplished, another transformation to the working space (e.g. Adobe RGB) can be performed. The details are described by [a href=\"http://www.poynton.com/notes/colour_and_gamma/ColorFAQ.html#RTFToC18]Poynton[/url]. What is the difference?

As Bruce Fraser explained in Real World Photoshop CS2, a custom RGB space contains there elements: Gamma, White Point, and Primaries.

In the raw file some of these are implicit. The gamma is one, the white point is described by the white balance data in the raw file, and the primaries are described in the matrix. What is missing?
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9315



WWW
« Reply #9 on: November 23, 2007, 09:16:56 AM »
ReplyReply

Quote
In the raw file some of these are implicit. The gamma is one, the white point is described by the white balance data in the raw file, and the primaries are described in the matrix. What is missing?
[a href=\"index.php?act=findpost&pid=155205\"][{POST_SNAPBACK}][/a]

The spectral sensitivity of the chip.

Bruce's book points out that these chips are simply counting photons. There are colored filters over the matrix (we don't know anything about those colored filters). It is this reason I said the Raw file is Grayscale but I should probably have said something like the Raw file is essentially Grayscale.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #10 on: November 23, 2007, 10:14:21 AM »
ReplyReply

Quote
The spectral sensitivity of the chip.

Bruce's book points out that these chips are simply counting photons. There are colored filters over the matrix (we don't know anything about those colored filters). It is this reason I said the Raw file is Grayscale but I should probably have said something like the Raw file is essentially Grayscale.
[a href=\"index.php?act=findpost&pid=155206\"][{POST_SNAPBACK}][/a]

Those matrix coefficients describe the spectral characteristics of the chip; again, without this information, it would not be possible to decode the raw file.

You state that a raw file is grayscale, but by the same token, as Bruce states on p. 119 of Real World PSCS2, "In Photoshop, files saved in the RGB mode typically uses a set of three 8 bit grayscale files..." [bold added for emphasis]. No essential difference here.
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9315



WWW
« Reply #11 on: November 23, 2007, 11:06:50 AM »
ReplyReply

Quote
Those matrix coefficients describe the spectral characteristics of the chip; again, without this information, it would not be possible to decode the raw file.

You state that a raw file is grayscale, but by the same token, as Bruce states on p. 119 of Real World PSCS2, "In Photoshop, files saved in the RGB mode typically uses a set of three 8 bit grayscale files..." [bold added for emphasis]. No essential difference here.
[a href=\"index.php?act=findpost&pid=155214\"][{POST_SNAPBACK}][/a]

Let's quote Bruce on a more recent book (and then we have the Schewe update which I haven't seen yet):

Page 2 of RW camera Raw:

Quote
A Raw file is a record of the sensor data, so lets look at what the sensor in a digital camera actually captures. A number of different technologies get lumped into the category of "digital camera." but virtually all the cameras supported by the Camera Raw plug-in are of a type known as "mosaic sensor" or "color filter array" cameras (virtually all because versions 2.2 and later support the Sigma camera based on the Foveon's X3 technology). The first point is that stripped-array Raw files are grayscale!

Color filter array cameras use a two dimensional area array to collect the photons that are recorded in the image. The array is made up of rows and columns of photosensitive detectors-typically CCD or COMS, to form the image.

Further on page 3 he writes:

Quote
But the sensors in the array, whether CCD or CMOS just count photons-they produce a charge proportional to the amount of light they receive-without recording any color information.[/u] The color information is produced by color filters that are applied over the individual elements in the array in a process known as "striping"-hence the term "striped array".

Getting back to your quote of Bruce's, its apples and oranges here. He's talking about an RGB file which of course is three Grayscale channels. That's not what he (or I) are talking about with respect to Raw. AFTER demoiscing, when the data is rendered it is in some RGB color space (in Camera Raw and LR, that's ProPhoto RGB linear encoded gamma) and then you can pick an output color space (an encoding color space). But long before this, your Raw data is Grayscale.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #12 on: November 23, 2007, 11:48:02 AM »
ReplyReply

Quote
Let's quote Bruce on a more recent book (and then we have the Schewe update which I haven't seen yet):

Page 2 of RW camera Raw:
Further on page 3 he writes:
Getting back to your quote of Bruce's, its apples and oranges here. He's talking about an RGB file which of course is three Grayscale channels. That's not what he (or I) are talking about with respect to Raw. AFTER demoiscing, when the data is rendered it is in some RGB color space (in Camera Raw and LR, that's ProPhoto RGB linear encoded gamma) and then you can pick an output color space (an encoding color space). But long before this, your Raw data is Grayscale.
[a href=\"index.php?act=findpost&pid=155228\"][{POST_SNAPBACK}][/a]

I would say that it is more like oranges and tangerines. Both raw and RGB are monochrome, but carry encoded color information. The typical Bayer CFA raw file has four components: green 1 , green 2, blue and red. An RGB file contains three components. What is the difference? You don't need to demosaic to view color information. If you decoded the four streams and viewed the result from a distance, the eye would blend the primaries into the proper colors.

So in essence, Bruce has stated in one publication that RGB consists of 3 monochrome components, and in another that raw has 4 monochrome components. I don't think that the principles of color theory have changed in the more current publication. Furthermore, what is your response to the Adobe DNG specification which refers to a camera color space? This may all be semantics, but the spaces are more similar than they are different. Three by three matrix transformations can be used on either.
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9315



WWW
« Reply #13 on: November 23, 2007, 11:55:25 AM »
ReplyReply

Quote
So in essence, Bruce has stated in one publication that RGB consists of 3 monochrome components, and in another that raw has 4 monochrome components. I don't think that the principles of color theory have changed in the more current publication.

So what's the color space of the raw data?
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #14 on: November 23, 2007, 03:29:40 PM »
ReplyReply

Quote
So what's the color space of the raw data?
[a href=\"index.php?act=findpost&pid=155236\"][{POST_SNAPBACK}][/a]

It is the native color space of the camera--in the quoted DCRaw case, that specified by the matrix values shown. In DCRaw, those are used to convert from camera to CIE XYZ. You can then use another 3 by 3 matrix conversion to your working space. Adobe DNG also describes a camera color space. How can you reasonably deny these facts? Did you look at the DNG specification?
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5542


WWW
« Reply #15 on: November 23, 2007, 03:43:16 PM »
ReplyReply

Quote
It is the native color space of the camera
[a href=\"index.php?act=findpost&pid=155288\"][{POST_SNAPBACK}][/a]


While it may be spec as a 3x3 matrix, what the "camera color space is" is actually the identification of the camera's spectral properties at a given white point...and as such is NOT a "profile" in ICC parlance...so while a camera DOES have a spectral response (at a given white point) is doesn't have a "color space" as it relates to a working space and it's also not an input profile...so saying that a camera does not have a color space profile attached to it would be correct.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #16 on: November 23, 2007, 04:21:23 PM »
ReplyReply

Quote
While it may be spec as a 3x3 matrix, what the "camera color space is" is actually the identification of the camera's spectral properties at a given white point...and as such is NOT a "profile" in ICC parlance...so while a camera DOES have a spectral response (at a given white point) is doesn't have a "color space" as it relates to a working space and it's also not an input profile...so saying that a camera does not have a color space profile attached to it would be correct.
[a href=\"index.php?act=findpost&pid=155294\"][{POST_SNAPBACK}][/a]

ICC or not, then why does Adobe refer to the Camera Color Space in the DNG specification? Someone at Adobe must consider it to be a color space. As mentioned above, it does have the three main elements of an ICC matrix color space (primaries, TRC, and white point), and it can be  transformed to an ICC working space via a 3 by 3 matrix transform. If it looks like a duck, walks like a duck, and quacks like a duck, it may be duck or at least something pretty close. A document in aRGB may not have a color space tag attached to it, but one can be assigned.
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5542


WWW
« Reply #17 on: November 23, 2007, 04:32:06 PM »
ReplyReply

Quote
ICC or not, then why does Adobe refer to the Camera Color Space in the DNG specification? Someone at Adobe must consider it to be a color space. As mentioned above, it does have the three main elements of an ICC matrix color space (primaries, TRC, and white point),
[a href=\"index.php?act=findpost&pid=155305\"][{POST_SNAPBACK}][/a]

It doesn't have a white point designation until AFTER conversion from the camera color (which is a more accurate way of describing what the raw file contains) to a "color space". All the camera color has is the chromaticities of the spectral response and the assumption of linear gamma. Without the designated white point (to be determined by the metadata tag or overridden by the processing software) it's not yet a complete "color space". Reading the spec is useful, but the terminology must be understood for what it means. This is the same limitation raw captures have when trying to apply input profiles (ala Capture One) to raw captures...since a raw capture doesn't yet have a complete set of chromaticities, gamma and white point, you can't really presume it has a profilable color color space.

And, returning to the OP question;
 
Quote
In a nut shell does a wider colour space mean a more noise in an image?

No...the noise in an image is dictated by the ISO settings in combination with the expansion of darker tones to lighter tone (lightening the shadows) such as would occur when moving the Exposure setting (or Brightness) to the plus values...
« Last Edit: November 23, 2007, 04:36:42 PM by Schewe » Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2882



« Reply #18 on: November 23, 2007, 05:00:24 PM »
ReplyReply

Quote
It doesn't have a white point designation until AFTER conversion from the camera color (which is a more accurate way of describing what the raw file contains) to a "color space".
[a href=\"index.php?act=findpost&pid=155308\"][{POST_SNAPBACK}][/a]

It does have white balance information, which DNG encodes as AsShotNeutral and as AsShotXY. Doesn't that count as white point designation?
Logged
Schewe
Sr. Member
****
Offline Offline

Posts: 5542


WWW
« Reply #19 on: November 23, 2007, 05:56:42 PM »
ReplyReply

Quote
It does have white balance information, which DNG encodes as AsShotNeutral and as AsShotXY. Doesn't that count as white point designation?
[a href=\"index.php?act=findpost&pid=155321\"][{POST_SNAPBACK}][/a]


You keep fishing for a raw file or DNG to have a "color space" as it is commonly associated with ICC style profiles and Photoshop working spaces. If you want to parse the words of the DNG spec and presume to attribute terms and their meanings, more power to ya, but it does a disservice to the industry to engage in a practice that will end up confusing people's understanding of raw linear captures and Photoshop working spaces and input profiles associated with scanners and cameras...

Raw captures do NOT have a "color space as it is defined by the ICC spec and associated with working color spaces in Photoshop". Raw captures are actually grayscale files whose color attributes are not yet assigned until the demosiacing process.

There, ya happy?

Ya know, sometimes you end up beating a dead horse bud...and the horse is beyond caring.
(and other people's eyes start to glaze over)
Logged
Pages: [1] 2 3 ... 6 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad