SD9 out-of-gamut Colors vs. the embedded Input Profile

[xpatUSA]

In RawTherapee and other Editors it is possible to change the said Profile.

I compared the standard sRGB to one randomly selected from RT's list

sRGB:

Random:

Then I opened these two in ColorThink along with ColorThink's 'Demo' sRGB:

The Cyan dots are from the Standard sRGB input profile - observe that there are five colors well outside of the sRGB triangle. Observe also that the 'random' profile green dots are all inside the triangle but do notice that many are bang up against the sides of the triangle - indicating the dreaded gamut or saturation-clipping.

And, finally, the white rogue dot that appears at the bottom on the x-axis is perhaps due to a bad pixel or two on the sensor or perhaps not.

I believe that a) just sticking with whatever profile your system embeds automatically is not necessarily a Good Thing and that b) pedants are repaid by going much deeper than "how it looks on the monitor or in the print".

ColorThink ... don't go home without it ...

 

[ole kekkonen]

Hi Ted,how do you use the color passport? Thinking about getting one and do I need the color think software, how does that work?

sorry for the stupid questions

 

[xpatUSA]

Hi Ted,how do you use the color passport?

My X-rite mini-checker came with a table of color values for each patch. I use the B,G,R,Y,M,C patches and I refer to the CIELAB values highlighted here:

I take a picture of the card; in this thread, the picture was taken outside in daylight - approximately CIE illuminant D65. In this case, I could checked the patch colors with the RGB color picker i.e. for blue compared the the color picker R,G,B values with the above listed 56,61,150. On my monitor today, 55,72,163, fow what it's worth.

For a deeper analysis, I prefer to use the CIELAB value and adjust the mid-gray patch in the image so that is truly neutral and approx 50 lightness before color-picking.

Thinking about getting one and do I need the color think software, how does that work?

ColorThink is only necessary if you are checking image or ICC profile gamuts. As to how that works, it is too big a subject to post in this forum.

sorry for the stupid questions

Not stupid at all, Ole.

Here's one for you to compare with the above list:

--

What you got is not what you saw.

 

[Scottelly]

Interesting post Ted. I think the color squares look good in both photos, though the original version from the camera look much more earth tone, and the color squares from that "other profile" look a lot brighter and more vivid (almost like its Landscape mode in a Quattro). I find it interesting that some of the data fell outside the sRGB color triangle. I wonder if Sigma made the camera to just ignore sRGB when using that color space, or what.

 

[Scottelly]

I still haven't bought a color checker.

 

[ole kekkonen]

Hi Ted,how do you use the color passport?

My X-rite mini-checker came with a table of color values for each patch. I use the B,G,R,Y,M,C patches and I refer to the CIELAB values highlighted here:

I take a picture of the card; in this thread, the picture was taken outside in daylight - approximately CIE illuminant D65. In this case, I could checked the patch colors with the RGB color picker i.e. for blue compared the the color picker R,G,B values with the above listed 56,61,150. On my monitor today, 55,72,163, fow what it's worth.

For a deeper analysis, I prefer to use the CIELAB value and adjust the mid-gray patch in the image so that is truly neutral and approx 50 lightness before color-picking.

Thinking about getting one and do I need the color think software, how does that work?

ColorThink is only necessary if you are checking image or ICC profile gamuts. As to how that works, it is too big a subject to post in this forum.

sorry for the stupid questions

Not stupid at all, Ole.

Here's one for you to compare with the above list:

Thanks a lot Ted for your time explaining me,I was not sure how to compare the image with the above list but the colors looks accurate and pleasing.

here's a. shot with a custom WB,I wonder if it looks like what I saw?

If I had a write mini checker I could take a photo of it in the field and compare it on the LCD screen with the checker?

P.s I read your post about 10 times, new terms new stupid questions arise LOL

 

[xpatUSA]

Hi Ted,how do you use the color passport?

My X-rite mini-checker came with a table of color values for each patch. I use the B,G,R,Y,M,C patches and I refer to the CIELAB values highlighted here:

I take a picture of the card; in this thread, the picture was taken outside in daylight - approximately CIE illuminant D65. In this case, I could checked the patch colors with the RGB color picker i.e. for blue compared the the color picker R,G,B values with the above listed 56,61,150. On my monitor today, 55,72,163, fow what it's worth.

For a deeper analysis, I prefer to use the CIELAB value and adjust the mid-gray patch in the image so that is truly neutral and approx 50 lightness before color-picking.

Thinking about getting one and do I need the color think software, how does that work?

ColorThink is only necessary if you are checking image or ICC profile gamuts. As to how that works, it is too big a subject to post in this forum.

sorry for the stupid questions

Not stupid at all, Ole.

Here's one for you to compare with the above list:

Thanks a lot Ted for your time explaining me,I was not sure how to compare the image with the above list but the colors looks accurate and pleasing.

Ole, how those colors look on your your screen do not tell you how accurate they are; and, sorry, but how pleasing they look is not relevant to an accuracy test.

here's a. shot with a custom WB,I wonder if it looks like what I saw?

If I had a write mini checker I could take a photo of it in the field and compare it on the LCD screen with the checker?

P.s I read your post about 10 times, new terms new stupid questions arise LOL

Let's get down to the basis of a camera color accuracy test:

1) shoot something which has a known color or colors. By "known", that something is an artifact which has a published value. The published value should be given for a certain lighting e.g. daylight, warm light, cool light, da-di-daa ...

Some people do not like SD9 yellows. In the table above, my yellow patch is RGB = 231, 199, 31

(note: the patch is not pure yellow. pure yellow is 255, 255, 0.)

2) open the image and use a color picker to measure the something's color - make an note.

On my computer, the color picker says RGB = 223, 187, 0.

3) compare.

Comparing RGBs is not intuitive. We are talking about yellows so the main interest is the the red and greens and their ratios red/green ... 1.16 for the table value and 1.19 for my color picker. Almost the same but with the image yellow tending a little more towards red.

Pure yellow is a ratio of 1.00 ... more is towards red, less is towards green.

The blue is a different story! At zero compared to 31, the image yellow is 100% saturated ... not good for skin color, as has been said here before.

RGB is a 3D color model and is not easy to translate into how we see color. An easier model is Hue, Saturation, Brightness (HSB aka HSV). For HSB my color picker says 50°, 100%, 88%. In HSB the hue for pure yellow is 60° ... less is towards red.

HSL and HSV - Wikipedia

en.wikipedia.org

Maybe my image is more pleasing because some or even all the colors are more saturated than the published table values ...

Questions are welcome!

--
What you got is not what you saw.

 

[ole kekkonen]

Hi Ted,how do you use the color passport?

My X-rite mini-checker came with a table of color values for each patch. I use the B,G,R,Y,M,C patches and I refer to the CIELAB values highlighted here:

I take a picture of the card; in this thread, the picture was taken outside in daylight - approximately CIE illuminant D65. In this case, I could checked the patch colors with the RGB color picker i.e. for blue compared the the color picker R,G,B values with the above listed 56,61,150. On my monitor today, 55,72,163, fow what it's worth.

For a deeper analysis, I prefer to use the CIELAB value and adjust the mid-gray patch in the image so that is truly neutral and approx 50 lightness before color-picking.

Thinking about getting one and do I need the color think software, how does that work?

ColorThink is only necessary if you are checking image or ICC profile gamuts. As to how that works, it is too big a subject to post in this forum.

sorry for the stupid questions

Not stupid at all, Ole.

Here's one for you to compare with the above list:

Thanks a lot Ted for your time explaining me,I was not sure how to compare the image with the above list but the colors looks accurate and pleasing.

Ole, how those colors look on your your screen do not tell you how accurate they are; and, sorry, but how pleasing they look is not relevant to an accuracy test.

here's a. shot with a custom WB,I wonder if it looks like what I saw?

If I had a write mini checker I could take a photo of it in the field and compare it on the LCD screen with the checker?

P.s I read your post about 10 times, new terms new stupid questions arise LOL

Let's get down to the basis of a camera color accuracy test:

) shoot something which has a known color or colors. By "known", that something is an artifact which has a published value. The published value should be given for a certain lighting e.g. daylight, warm light, cool light, da-di-daa ...

Some people do not like SD9 yellows. In the table above, my yellow patch is RGB = 231, 199, 31

(note: the patch is not pure yellow. pure yellow is 255, 255, 0.)

2) open the image and use a color picker to measure the something's color - make an note.

On my computer, the color picker says RGB = 223, 187, 0.

3) compare.

Comparing RGBs is not intuitive. We are talking about yellows so the main interest is the the red and greens and their ratios red/green ... 1.16 for the table value and 1.19 for my color picker. Almost the same but with the image yellow tending a little more towards red.

Pure yellow is a ratio of 1.00 ... more is towards red, less is towards green.

The blue is a different story! At zero compared to 31, the image yellow is 100% saturated ... not good for skin color, as has been said here before.

RGB is a 3D color model and is not easy to translate into how we see color. An easier model is Hue, Saturation, Brightness (HSB aka HSV). For HSB my color picker says 50°, 100%, 88%. In HSB the hue for pure yellow is 60° ... less is towards red.

https://en.wikipedia.org/wiki/HSL_and_HSV

Maybe my image is more pleasing because some or even all the colors are more saturated than the published table values ...

Questions are welcome!

Thank you!