Update on the vanishing blue.

[george013]

and as requested: SOOC. In C1 the petals show no yellow and yellow shows no blue

Just with IrfanView. The full sooc Average blue 22.11


And a crop of the petals. Average blue 2.36


I don't know what you mean with yellow. Only R,G and B counts.


George

[Manfred M]

Brian - part of the reason I mentioned this "issue" in the first place was a bit of a follow up to another recent conversation we had on shooting raw. In that discussion I mentioned that most experienced photographers would recommend capturing a scene in such a way as to maximize the data quality. Once that has been done, the photographer has a very solid foundation to build the image on. Highlight clipping (values of 256, 256, 256) or crushed shadow detail (values of 0, 0, 0) are undesirable as this restricts what can be done with the image in edit. When this happens at both ends of the histogram, we refer to that situation as the scene having exceeded the dynamic range of the camera's sensor. If this is not the case, the photographer may have some things he or she can do to reduce or eliminate blocking or clipping by managing the exposure. Understanding the histogram and what does and does not matter in the final image should be used to make decisions regarding the specific image.

As an example small areas of specular highlights or bright light sources can generally be allowed to clip without hurting the final image. Likewise small areas where shadow details are blocked will definitely look better in the final image that large areas of blown out highlights.

We sometimes see situations where the we don't get totally blown highlights or blocked shadow details. Sometimes only one (or two) of the colour channels show that highlight or shadow detail have been lost. That is what we see in this image, Brian. While Capture One has the tools to see this, I don't think they are as good as the ones that Adobe provides to visualize the extent of the "issue".




Looking at all three colour channels at the same time suggests two things to me:

1. There is very little blue in the image (not surprising when we look at it) and the little there is is mostly crushed (0 value).

2. The image is underexposed as the tiny areas where there are specular highlights show in the light gray areas, so I would suggest that the image is underexposed. Improving the expsure (by not dialing in -0.3 EC as per your original posting) would have been a move in the right direction. Clipping the specular highlights (there are very few of these, to protect the shadow detail) would have been a good choice.

A second feature in the Adobe software, that I have not been able to replicate in Capture One shows the extent of where the blue channel issues are (highlighted in yellow). There are also some areas of magenta, which shows that there are certain areas where the green channel (complement of magenta) is clipped. The problem with this functionality is that it shows you the affected areas, but not the extent of how far the colour is away from the original.




Ted (and others) are quite right that this is not a condition that we normally see in our images as in nature pure colours don't really exist. We sometimes see them in certain light sources, so if you take pictures of "neon lights", this can occur.

This brings us to the next part of the question and that is the ability of our cameras to capture the colours correctly. Nicely said, a decent modern camera can capture any wavelength of light that humans can see (and more). This is data and has to be converted into an image format before we can see and manipulate. This is something that the raw converter, whether it is an Adobe product like Lightroom or Camera Raw, a Phase One product like Capture One, etc. do for us. One of the attributes required of an image file is to assign it to a Colour Space, which tells our editing software how to interpret the specific R, G and B values. In photography we tend to use one of three colour spaces; sRGB which can reproduce around 1/3 of the colours humans can see, AdobeRGB which can reproduce around 2/3 of the colours we can see and ProPhoto RGB, which can reproduce about 80% of the colours we can see. The sRGB colour space is the defacto standard for any images posted on the internet. The other colour spaces can handle more colours, but it is important to remember that these tend to be the more vibrant shades, and these tend to the ones that we generally encounter less often in photography. It is easy to find computer screens that handle 100% of the sRGB colour space. There are a few, more expensive ones, that can virtually all of the AdobeRGB colour space. There are none that come close to handling the colours that ProPhoto can reproduce.

Colours that cannot be reproduced by the colour space or our output devices are referred to as being Out-of-Gamut (OOG). When that happens our system software uses something called a Rendering Intent to remap the colours to something that can be displayed. The most common rendering intent that we use is called Relative Colorimetric and what it does it remaps the out of gamut colour to the nearest one that can be displayed. If you work in a wider gamut colour space and convert it into, say sRGB there can be colours that come out of the process that behave much like the ones we see here. The closest colour that can be reproduced in the colour space could have a "0" value in the blue channel.

The final issue is what to do about it. Clearly, avoiding the problem by taking care during capture and PP work would be the place to start. That is not always possible and here a "do nothing" solution can work or doing what Ted suggests, working the image by changing the brightness or the saturation might be an acceptable step, depending on whether it gives you a result you can live with. I have run into issues, especially when printing, were the results are simply not where I want them to be, and in these cases, there are times where I abandoned a project because I did not like the look I wanted. Perhaps when they come out with a ProPhoto RGB computer screen and more vivid papers and ink sets, I might go back to them.

[ccphoto]

and as requested: SOOC. In C1 the petals show no yellow and yellow shows no blue

According to my color picker in Photoshop, there is some blue in your yellows in the RAW file (I picked the petal immediately to the right of the Y in the CMYK color file.)





After a Color Balance Adjustment and a bit of contrast added, the blue is gone.



My final edit and I am sure since I am not there to observe the on site colors, it might differ from your own



Perhaps this does address all the color charting and math associated with this thread, but it does address eliminating the blue from the yellow and I think that was the original intent.

[Manfred M]

I don't know what you mean with yellow. Only R,G and B counts.

George - remember colour theory and complementary colours. Look back at what I posted in #12. Every time you change the blue value, you affect the yellow value.

[george013]

But I can only count R,G and B.

George

[Manfred M]

But I can only count R,G and B.

George

That's correct. But any additive colour can be created from these three primary colours. If you want yellow, turn the red and green channels all the way up to a value of 255 and turn the blue channel to 0.

[george013]

Why are you (deliberately?) restricting your posts to the RGB color model?

In my editor, the color-picker displays RGB, HSV and CIELAB all at same time. Each set of values has it's use(s).

That's where this thread was about, vanishing blue. It seems simple and direct answers are impossible.

George

[Manfred M]

That's where this thread was about, vanishing blue. It seems simple and direct answers are impossible.

George

If we restrict ourselves to the most common approaches used by most photographers, it does not have to be at all that complicated. In my view, it is important that photographers have a basic understanding of at least the RGB colour model and the three colour spaces that are commonly used in photography.

It's challenging to edit in colour without understanding the basics of the RGB colour model.

[JBW]

How did you get the SOOC? Did you shoot raw, and if so, how did you produce the image you posted? I'm still trying to figure out where that degree of saturation was introduced.

BTW, although I would crop it a bit, I much prefer the uncropped to the original. The original shows a few fine details but is mostly featureless, out-of-focus yellow. The original is a much more interesting photo, IMHO.

How did I get the SOOC? About five years ago M'Lady bought me a FujiFilm S4200. Fast forwarding... I lined up the camera, pressed my auto shutter release and took the shot using my Sony Alpha a68 connected to my Tamron 90mm 272E macro lens. Natural light.

I always shoot in Raw, usually full manual. Except for the ISO which stays at 100 and White Balance which I set to auto.

I came back into the house, took out my memory card, slipped it into the computer ( a very nice and custom built computer) and transferred directly into Capture 1 Sony Pro V11.1.2.

Then I scaled it down, shipped it out as a JPEG at 100% uploaded to my WordPress blog, copied the image location and posted it to CiC.

Hope that helps.

I agree the petals can make for a nice shot but hardly a macro. I was going for that one perfectly focused area rendered in maximum detail

[JBW]

This brings us to the next part of the question and that is the ability of our cameras to capture the colours correctly. Nicely said, a decent modern camera can capture any wavelength of light that humans can see (and more). This is data and has to be converted into an image format before we can see and manipulate. This is something that the raw converter, whether it is an Adobe product like Lightroom or Camera Raw, a Phase One product like Capture One, etc. do for us. One of the attributes required of an image file is to assign it to a Colour Space, which tells our editing software how to interpret the specific R, G and B values. In photography we tend to use one of three colour spaces; sRGB which can reproduce around 1/3 of the colours humans can see, AdobeRGB which can reproduce around 2/3 of the colours we can see and ProPhoto RGB, which can reproduce about 80% of the colours we can see. The sRGB colour space is the defacto standard for any images posted on the internet. The other colour spaces can handle more colours, but it is important to remember that these tend to be the more vibrant shades, and these tend to the ones that we generally encounter less often in photography. It is easy to find computer screens that handle 100% of the sRGB colour space. There are a few, more expensive ones, that can virtually all of the AdobeRGB colour space. There are none that come close to handling the colours that ProPhoto can reproduce.

Colours that cannot be reproduced by the colour space or our output devices are referred to as being Out-of-Gamut (OOG). When that happens our system software uses something called a Rendering Intent to remap the colours to something that can be displayed. The most common rendering intent that we use is called Relative Colorimetric and what it does it remaps the out of gamut colour to the nearest one that can be displayed. If you work in a wider gamut colour space and convert it into, say sRGB there can be colours that come out of the process that behave much like the ones we see here. The closest colour that can be reproduced in the colour space could have a "0" value in the blue channel.
go back to them.

This may be where part of the solution lies.

I have my caera set to sRGB and capture 1 to sRGB 19662.1

I alos have C1 set to Perceptual in the color options. Would it be better if I set it to Relative or Absolute colorimtric or perhaps Saturation? Do i stay with the standard color wheel or switch to Victorscope?

Brian

[JBW]

If we restrict ourselves to the most common approaches used by most photographers, it does not have to be at all that complicated. In my view, it is important that photographers have a basic understanding of at least the RGB colour model and the three colour spaces that are commonly used in photography.

It's challenging to edit in colour without understanding the basics of the RGB colour model.

Not so challenging most of the time. The lizard was quite easy. It does get more challenging as the extremes are approached. I'll go out when I can and try this shot using the dark zero and the bright 0 to see the difference. And have some fun with the color settings in C1.

[Manfred M]

This may be where part of the solution lies.

I have my caera set to sRGB and capture 1 to sRGB 19662.1

I alos have C1 set to Perceptual in the color options. Would it be better if I set it to Relative or Absolute colorimtric or perhaps Saturation? Do i stay with the standard color wheel or switch to Victorscope?

Brian

Typically most photographers will set the default to Relative Colorimetric but will check to see if the image looks better in Perceptual. Absolute and Saturation are used in graphic art work, not photography.

In terms of how you pick colours, that is personal preference. People coming from the video side of things tend to prefer the Vectorscope approach as it is the default approach in professional non-linear video editing software. It makes no difference which one you use.

Generally people who shoot in raw do not use sRGB as their working colour space as it only covers about 1/3 of the colours that humans can see. The general direction is to use a wide gamut colour space for editing work and convert to sRGB before posting to the internet. The trend is to work in ProPhoto RGB although I know some photographers who do use AdobeRGB from time to time, but this is more related to people who prepare work for physical output to photo printers or to commercial printing. presses.

[xpatUSA]

This may be where part of the solution lies.

I have my camera set to sRGB and Capture 1 to sRGB 19662.1

I alos have C1 set to Perceptual in the color options. Would it be better if I set it to Relative or Absolute colorimetric or perhaps Saturation? Do i stay with the standard color wheel or switch to Victorscope?

Brian

We have to be careful about "Perceptual". If the embedded ICC profile in your image file is of the matrix type (no color look-up tables, CLUTs) the CMS in the rendering application will use Relative Colorimetric in spite of what it says in the profile or what you selected in C1. Most display profiles are of the matrix type, so selecting Perceptual does nothing.

Most printer profiles have CLUTs (tagged A2B0, etc) and Perceptual will work (act like the CiC tutorial) with those.

Looking in where the profiles live on your computer hard drive, matrix profiles are about 5-10K; profiles with CLUTs are about 200K or so.

The most common sRGB profile IEC61966-2.1 is of the matrix type and will not give a Perceptual rendering. For posting or viewing on your computer I'd stick with IEC61966-2.1. You can leave C1 set to Perpetual; it will make no difference unless you print.

I have no idea what Victorscope is, sorry.

[DanK]

I have my caera set to sRGB and capture 1 to sRGB 19662.1

I alos have C1 set to Perceptual in the color options.

Since you are shooting raw, the sRGB setting on the camera should have no effect on the file. However, I don't know C1. Does it impose the color space set in the camera? Lightroom doesn't. Assuming it doesn't, it would seem that the answer most likely lies in something else C1 is doing in its initial rendering (since Brian isn't doing anything else).

Brian, if you have a way to upload the raw file, e.g., a dropbox location, I would be happy to read it into Lightroom and see what happens.

PS: PM me if you want to do this

[xpatUSA]

Brian, if you have a way to upload the raw file, e.g., a dropbox location, I would be happy to read it into Lightroom and see what happens.

PS: PM me if you want to do this

I would be interested in the raw, too. Would like to open it in RawDigger and also RawTherapee to see what all the fuss is about!

[Manfred M]

We have to be careful about "Perceptual". If the embedded ICC profile in your image file is of the matrix type (no color look-up tables, CLUTs) the CMS in the rendering application will use Relative Colorimetric in spite of what it says in the profile or what you selected in C1. Most display profiles are of the matrix type, so selecting Perceptual does nothing.

Most printer profiles have CLUTs (tagged A2B0, etc) and Perceptual will work (act like the CiC tutorial) with those.

Looking in where the profiles live on your computer hard drive, matrix profiles are about 5-10K; profiles with CLUTs are about 200K or so.

The most common sRGB profile IEC61966-2.1 is of the matrix type and will not give a Perceptual rendering. For posting or viewing on your computer I'd stick with IEC61966-2.1. You can leave C1 set to Perpetual; it will make no difference unless you print.

I have no idea what Victorscope is, sorry.

Ted - I am going to have to somewhat disagree with you on this one, at least so far as how Photoshop (and Lightroom) have implemented this functionality. Both have a mode referred to as "Soft Proofing" where the user can set the software to emulate what output will look like for any ICC profile loaded on the computer. Normally this is used to emulate printer / paper combinations for preparing output to print and the impact of how the rendering intents apply to the specific image are definitely part of this functionality. I can definitely see how perceptual versus relative colorimetric work on the computer screen. This mode is very helpful when preparing for print output and I will even make paper selections based on what I see on screen and will decide on which rendering intent to use for printing. Being an emulation, it is not a 100% solution, but certainly close to it.

Being a generic operating mode, I can also dial it in for colour spaces icc profiles and can confirm that the screen output looks different from the default view to what it looks like when I toggle it on and off for a specific colour space profile. What is less clear is if changing the rendering intent selection has any impact on the screen output. If there is, I would suggest it very subtle.

[xpatUSA]

Ted - I am going to have to somewhat disagree with you on this one, at least so far as how Photoshop (and Lightroom) have implemented this functionality. Both have a mode referred to as "Soft Proofing" where the user can set the software to emulate what output will look like for any ICC profile loaded on the computer. Normally this is used to emulate printer / paper combinations for preparing output to print and the impact of how the rendering intents apply to the specific image are definitely part of this functionality. I can definitely see how perceptual versus relative colorimetric work on the computer screen. This mode is very helpful when preparing for print output and I will even make paper selections based on what I see on screen and will decide on which rendering intent to use for printing. Being an emulation, it is not a 100% solution, but certainly close to it.

Being a generic operating mode, I can also dial it in for colour spaces icc profiles and can confirm that the screen output looks different from the default view to what it looks like when I toggle it on and off for a specific colour space profile. What is less clear is if changing the rendering intent selection has any impact on the screen output. If there is, I would suggest it very subtle.

Thanks for the information re: soft-proofing. I was trying to keep it simple and was not referring to that at all, but I should have made that clear. My post applies to using a browser or a viewer to render an image file, per your last two sentences above.. Somewhere on the Adobe site the business of matrix/perceptual gets a mention, perhaps by Chan, Schew or Fraser - my memory fails me, unfortunately.

Bob Johnson confirms what I was trying to say far better than myself:

http://www.earthboundlight.com/photo...erceptual.html

The knowledgeable Elle Stone mentions it right in the header in here, too:

https://ninedegreesbelow.com/photogr...n-intents.html

Perhaps in soft-proofing the target print-profile gets used which is almost certain to have CLUTs? What do you say?

[DanK]

Here's some additional data, thanks to Brian's uploading the raw file.

I imported it into Lightroom, using a variety of different import profiles. There is blue in all of the yellow part of the flower, although the amounts do differ depending on the profile. Using the current standard ("Adobe color"), most of the yellow area has blue values centered around 20 and mostly between 15 and 25, although a few areas go below 10. I didn't notice any zeros. There are modest changes if I apply other rendering profiles. For example, the "vivid" profile, which I never use but which I assume boosts saturation, reduces blue values, but nowhere nearly to zero.

Then I took the default Adobe color version and exported it as a JPEG at both 100% and 92% quality. Regardless of the quality setting, the result was a lot like Brian's initial posting: lots of zero blue values and other values that were very close, e.g., <5.

So does this mean that blues in the original image were outside the sRGB gamut?

Here's one of my own flowers that is informative:



This was about the yellowest flower I could find in my shots. It is a stack of several individual shots. All of the original shots had very low blue values in the petals, in many areas between 7 and 9, but few smaller than that and no zeros. Compositing in Zerene didn't change much, but editing in LR (some +clarity, +vibrance, and a bit of contrast via a curve) did reduce the blues somewhat. Editing in PS (levels, curves again, sharpening) decreased blues even more, to 2 and 3. Exporting to JPEG in this case also reduced many of the blues to zero.

[xpatUSA]

Here's some additional data, thanks to Brian's uploading the raw file.

I imported it into Lightroom, using a variety of different import profiles. There is blue in all of the yellow part of the flower, although the amounts do differ depending on the profile. Using the current standard ("Adobe color"), most of the yellow area has blue values centered around 20 and mostly between 15 and 25, although a few areas go below 10. I didn't notice any zeros. There are modest changes if I apply other rendering profiles. For example, the "vivid" profile, which I never use but which I assume boosts saturation, reduces blue values, but nowhere nearly to zero.

Then I took the default Adobe color version and exported it as a JPEG at both 100% and 92% quality. Regardless of the quality setting, the result was a lot like Brian's initial posting: lots of zero blue values and other values that were very close, e.g., <5.

So does this mean that blues in the original image were outside the sRGB gamut?

Usually Dan, it means that the opposing color (in the CIE 1931 gamut diagram) to blue (which is yellow) was out of sRGB gamut; the conversion to sRGB forces the out of gamut yellow to the sRGB gamut boundary which has the effect of pushing the blue lower in value, even to the point of bottoming at zero.

[DanK]

Usually Dan, it means that the opposing color (in the CIE 1931 gamut diagram) to blue (which is yellow) was out of sRGB gamut; the conversion to sRGB forces the out of gamut yellow to the sRGB gamut boundary which has the effect of pushing the blue lower in value, even to the point of bottoming at zero.

That makes sense. Thanks.

I found the smaller effects of edits intriguing. A simple curve, without luminance-only blending, increases saturation. So does vibrance, although it supposedly doesn't affect already-highly-saturated areas. But if the edits are increasing saturation, wouldn't that also push down the blues in the yellow areas?