Spectra of natural colours

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."
J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I have a CGATs file Iliah was kind enough to produce if you so desire. Then one can (as I did) plot the colors gamut in something like ColorThink Pro or any software you have that can plot a color list.

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors. I generated a few graphs of various green leaves and will post them later.

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors.

Click to expand...

Telling.

J A C S said:

I generated a few graphs of various green leaves and will post them later.

Click to expand...

But not all 218 reflectance spectra of natural plant flower color, where it’s been shown, some orange and yellow samples are clipped in sRGB:



Again, telling.

digidog said:

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors.

Click to expand...

Telling.

Member said:

I generated a few graphs of various green leaves and will post them later.

Click to expand...

But not all 218 reflectance spectra of natural plant flower color, where it’s been shown, some orange and yellow samples are clipped in sRGB:

https://www.dpreview.com/forums/post/59559368

https://www.dpreview.com/forums/post/59559609

Again, telling.

Click to expand...

Stop behaving as a donkey-hole, grow up. I was talking about graphing the spectra, there is no sRGB there. There are too many spectra, my choice of green leaves is a personal preference.

I have to put you in my ignore file now, your behavior is too childish and you miss my point almost all the time. Bye.

J A C S said:

digidog said:

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors.

Click to expand...

Telling.

J A C S said:

I generated a few graphs of various green leaves and will post them later.

Click to expand...

But not all 218 reflectance spectra of natural plant flower color, where it’s been shown, some orange and yellow samples are clipped in sRGB:

https://www.dpreview.com/forums/post/59559368

https://www.dpreview.com/forums/post/59559609

Again, telling.

Click to expand...

Stop behaving as a donkey-hole, grow up. I was talking about graphing the spectra, there is no sRGB there.

Click to expand...

There is sRGB to compare the 218 coloured samples collected from the nature against.

From the site with the data some of us fully examining:

The reflectance spectra of 218 coloured samples collected from the nature. Samples include flowers and leaves and other colourful plants. Each spectrum consists of 61 elements that are raw data got from the output of the 12 bit A/D-converter of the AOTF color measuring equipment. All the element values should be between 0 and 4096. However there can exist some values which are larger than 4096. Those element values should be corrected to the value 4096.

J A C S said:

There are too many spectra, my choice of green leaves is a personal preference.

Click to expand...

There are too many for you and you’ve decided to plot a subset unlike others who were willing and able to examine all the data. Forgive me for asking about your predetermined testing methodology which is, by your own admission, going to be incomplete.

J A C S said:

I have to put you in my ignore file now**, your behavior is too childish and you miss my point almost all the time. Bye.

Click to expand...

Will not stop others and myself from calling you out.

**“Most men would rather deny a hard truth than face it.” -George R.R. Martin, A Game of Thrones

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors. I generated a few graphs of various green leaves and will post them later.

Click to expand...

Thanks.

I haven't used a spreadsheet program for decades, so I would have to mess around finding out how to generate a graph, perhaps in Libre Office.

I do wonder why they posted a straightforward small text file as a "tar.gz", But it didn't take long to unwrap it. (I use "Total Commander" for such tasks.)

D Cox said:

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors. I generated a few graphs of various green leaves and will post them later.

Click to expand...

Thanks.

I haven't used a spreadsheet program for decades, so I would have to mess around finding out how to generate a graph, perhaps in Libre Office.

Click to expand...

I don't understand what is the big deal regarding assessing the applicability of the colors from that link and sRGB. It was mentioned here in the original post:

https://www.dpreview.com/forums/post/59561972

it is just a few lines of code in Octave / Matlab / Python.

D Cox said:

I do wonder why they posted a straightforward small text file as a "tar.gz", But it didn't take long to unwrap it. (I use "Total Commander" for such tasks.)

Click to expand...

Why make it complicated? That is just a tar file.

Just use "tar -xvfz FILENAME" on command line in a terminal on Mac / Linux. If on Windows (who does scientific works on Windows in any case) then download Cygwin and then use that tar command.

--
Dj Joofa
http://www.djjoofa.com

My apology for semi-hijacking this follow-on thread, but I also had a question concerning a post made in the other thread that filled up. Digidog posted a comparison of Granger rainbow/charts constructed in ProPhoto and sRGB (8-bit for both). See here. The respective histograms showed considerable banding in the sRGB version but not in the ProPhoto version. My question to Digidog: How were the Granger rainbows constructed? I can't seem to duplicate the banding in the sRGB versions that I'm creating, so I'd be curious to learn more about the steps used by Digidog. I'm using Photoshop and assume the same was done by Digidog. Thanks, and again my apology for going off-topic in this specific thread. The question didn't seem worthy of a whole separate thread.

Joofa said:

D Cox said:

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors. I generated a few graphs of various green leaves and will post them later.

Click to expand...

Thanks.

I haven't used a spreadsheet program for decades, so I would have to mess around finding out how to generate a graph, perhaps in Libre Office.

Click to expand...

I don't understand what is the big deal regarding assessing the applicability of the colors from that link and sRGB. It was mentioned here in the original post:

https://www.dpreview.com/forums/post/59561972

it is just a few lines of code in Octave / Matlab / Python.

Click to expand...

I am more interested in the spectra than in colour spaces.

Joofa said:

Joofa said:

I do wonder why they posted a straightforward small text file as a "tar.gz", But it didn't take long to unwrap it. (I use "Total Commander" for such tasks.)

Click to expand...

Why make it complicated? That is just a tar file.

Just use "tar -xvfz FILENAME" on command line in a terminal on Mac / Linux. If on Windows (who does scientific works on Windows in any case) then download Cygwin and then use that tar command.

Click to expand...

That takes a lot longer than just clicking on the icon in Total Commander and copying the archive contents to where you want them.

But why not just post the text file as it is ? Why archive it ?

Not that it really matters.

knickerhawk said:

My apology for semi-hijacking this follow-on thread, but I also had a question concerning a post made in the other thread that filled up. Digidog posted a comparison of Granger rainbow/charts constructed in ProPhoto and sRGB (8-bit for both). See here. The respective histograms showed considerable banding in the sRGB version but not in the ProPhoto version. My question to Digidog: How were the Granger rainbows constructed? I can't seem to duplicate the banding in the sRGB versions that I'm creating, so I'd be curious to learn more about the steps used by Digidog. I'm using Photoshop and assume the same was done by Digidog. Thanks, and again my apology for going off-topic in this specific thread. The question didn't seem worthy of a whole separate thread.

Click to expand...

When working with 8-bits per channel, this can happen when converting from a wider gamut colorspace to a smaller gamut colorspace. The issue is that within a range of colors, the smaller gamut has more distinct device values. In some sense the smaller gamut space has a higher "precision" than the larger gamut space, but the precision of the data has already been reduced to the lower "precision" of the other space. Thus we end up with unused device values in our file.

This is one of the disadvantages of working in a larger colorspace when your final deliverable will be in a smaller colorspace.

You can avoid this problem by working in 16 bits per channel, and converting while still in 16 bits per channel.

When it comes to colorspaces, it's important to pay attention to both the theoretical and practical issues. Sometimes the practical issues will depend on which tools you are using.

Michael Fryd said:

knickerhawk said:

My apology for semi-hijacking this follow-on thread, but I also had a question concerning a post made in the other thread that filled up. Digidog posted a comparison of Granger rainbow/charts constructed in ProPhoto and sRGB (8-bit for both). See here. The respective histograms showed considerable banding in the sRGB version but not in the ProPhoto version. My question to Digidog: How were the Granger rainbows constructed? I can't seem to duplicate the banding in the sRGB versions that I'm creating, so I'd be curious to learn more about the steps used by Digidog. I'm using Photoshop and assume the same was done by Digidog. Thanks, and again my apology for going off-topic in this specific thread. The question didn't seem worthy of a whole separate thread.

Click to expand...

When working with 8-bits per channel, this can happen when converting from a wider gamut colorspace to a smaller gamut colorspace. The issue is that within a range of colors, the smaller gamut has more distinct device values. In some sense the smaller gamut space has a higher "precision" than the larger gamut space, but the precision of the data has already been reduced to the lower "precision" of the other space. Thus we end up with unused device values in our file.

This is one of the disadvantages of working in a larger colorspace when your final deliverable will be in a smaller colorspace.

You can avoid this problem by working in 16 bits per channel, and converting while still in 16 bits per channel.

When it comes to colorspaces, it's important to pay attention to both the theoretical and practical issues. Sometimes the practical issues will depend on which tools you are using.

Click to expand...

No offense intended, but I'm not interested in (nor require) your explanation of the differing effects of bit depth on varying color spaces. I'd simply like Digidog to explain the steps he used to construct his respective Granger rainbows. I'm curious as to why we seem to be getting different results.

knickerhawk said:

knickerhawk said:

...

Click to expand...

No offense intended, but I'm not interested in (nor require) your explanation of the differing effects of bit depth on varying color spaces. I'd simply like Digidog to explain the steps he used to construct his respective Granger rainbows. I'm curious as to why we seem to be getting different results.

Click to expand...

My apologies.

D Cox said:

Joofa said:

D Cox said:

J A C S said:

D Cox said:

I excerpt these bits from a current very long thread:

"The spectral measurements of natural colours available from University of Eastern Finland confirm that sRGB does not contain even the limited set of colours of natural objects they studied. Some orange and yellow samples are clipped in sRGB."

J A C S said:

"Thanks for the link, I was looking for something like this before. I downloaded the text file, imported it into an Excel spreadsheet and I am generating graphs now. I wish they did that (or maybe they did but I have not found it). I am more interested in eyeballing the spectral curves than computing coordinates in this and that space."

Did you manage to generate those graphs ?

If so I would very much like to see them.

Click to expand...

I generated a few. It was not easy (for me) because I had to reformat the text file by removing the wrapping, import into Excel, and fix some formatting errors. I generated a few graphs of various green leaves and will post them later.

Click to expand...

Thanks.

I haven't used a spreadsheet program for decades, so I would have to mess around finding out how to generate a graph, perhaps in Libre Office.

Click to expand...

I don't understand what is the big deal regarding assessing the applicability of the colors from that link and sRGB. It was mentioned here in the original post:

https://www.dpreview.com/forums/post/59561972

it is just a few lines of code in Octave / Matlab / Python.

Click to expand...

I am more interested in the spectra than in colour spaces.

Click to expand...

To me it appears that is even easier. The file already contains the spectra. Just untar it and plot it.

D Cox said:

D Cox said:

D Cox said:

I do wonder why they posted a straightforward small text file as a "tar.gz", But it didn't take long to unwrap it. (I use "Total Commander" for such tasks.)

Click to expand...

Why make it complicated? That is just a tar file.

Just use "tar -xvfz FILENAME" on command line in a terminal on Mac / Linux. If on Windows (who does scientific works on Windows in any case) then download Cygwin and then use that tar command.

Click to expand...

That takes a lot longer than just clicking on the icon in Total Commander and copying the archive contents to where you want them.

Click to expand...

My comment was meant to be taken more seriously, actually.

Linux / Mac / Windows with Cygwin offer a huge amount of easy, couple of lines (many times just one-liners) command line text processing suite of utilities that it is mind-boggling how any serious researcher would not use them? Indeed many (most??) do for cleaning and preparing text based data for experimentation. Even binary data many times.

D Cox said:

But why not just post the text file as it is ? Why archive it ?

Click to expand...

To compress and save space. It may be little in this case. But , still.

--
Dj Joofa
http://www.djjoofa.com

Michael Fryd said:

knickerhawk said:

Michael Fryd said:

...

Click to expand...

No offense intended, but I'm not interested in (nor require) your explanation of the differing effects of bit depth on varying color spaces. I'd simply like Digidog to explain the steps he used to construct his respective Granger rainbows. I'm curious as to why we seem to be getting different results.

Click to expand...

My apologies.

Click to expand...

None required, but thanks for the polite response.

Joofa said:

My comment was meant to be taken more seriously, actually.

Click to expand...

Really?

Joofa said:

Linux / Mac / Winodws with Cygwin offer a huge amount of easy, couple of lines (many times just one-liners) on command line text processing suite of utilities that it is mind-boggling how any serious researcher would not use them? Indeed many (most??) do for cleaning and preparing text based data for experimentation. Even binary data many times.

Click to expand...

I used Word with a few find/replace commands to reformat the text file into a form suitable for importing into Excel.

Do you really feel the need to discuss this? What is wrong with this forum?

The reason it is taking me so long is that (1) I have other things to do and (2) there are hundreds of spectra and I need to chose the ones I want to plot and to present them appropriately.

J A C S said:

Joofa said:

My comment was meant to be taken more seriously, actually.

Click to expand...

Really?

Click to expand...

Yes, really. Seriously.

J A C S said:

J A C S said:

Linux / Mac / Winodws with Cygwin offer a huge amount of easy, couple of lines (many times just one-liners) on command line text processing suite of utilities that it is mind-boggling how any serious researcher would not use them? Indeed many (most??) do for cleaning and preparing text based data for experimentation. Even binary data many times.

Click to expand...

I used Word with a few find/replace commands to reformat the text file into a form suitable for importing into Excel.

Click to expand...

Ha, if that is really what you do for data 'cleanup' before experimentation, then I guess my comment is really not meant for you.

J A C S said:

Do you really feel the need to discuss this? What is wrong with this forum?

Click to expand...

Do you really feel the need to comment? You are neither OP nor I replied to you. Just keep quiet and work on figuring out your work on Excel.

J A C S said:

The reason it is taking me so long is that (1) I have other things to do and (2) there are hundreds of spectra and I need to chose the ones I want to plot and to present them appropriately.

Click to expand...

Good luck!

--
Dj Joofa
http://www.djjoofa.com

knickerhawk said:

My apology for semi-hijacking this follow-on thread, but I also had a question concerning a post made in the other thread that filled up. Digidog posted a comparison of Granger rainbow/charts constructed in ProPhoto and sRGB (8-bit for both). See here. The respective histograms showed considerable banding in the sRGB version but not in the ProPhoto version. My question to Digidog: How were the Granger rainbows constructed?

Click to expand...

I can send you a few pages from my book tutorial on how one is built. But you can use any image as I later illustrated when called out for not using an actual image file. The results were the same!

knickerhawk said:

I can't seem to duplicate the banding in the sRGB versions that I'm creating, so I'd be curious to learn more about the steps used by Digidog.

Click to expand...

By editing the file! You can try all kinds of various edits on any image in any working space in 8-bits per color to produce banding. And that’s my point to Fryd; that picking on editing in ProPhoto RGB or any other RGB working space in 8-bits per color CAN produce banding. Depends on the edits, how many, what kind etc. As I said, for many, a big Duh moment.

He’s suggesting that ProPhoto RGB is in some way more dangerous to use because unless it’s in high bit, editing can cause banding. That’s somewhat true. And it’s also true of sRGB or any other RGB working space with that encoding. That’s WHY we have high bit capture and why we edit in high bit.

The ProPhoto FUD isn’t any more compelling than the sRGB FUD in terms of editing in 8-bits per color; it can show up.

For example, you do all the parametric edits in Lightroom or ACR and export a ProPhoto RGB JPEG in of course 8-bits per color. Not a lick of banding. So ProPhoto RGB is problematic? Of course not. Do the same with sRGB, same results. Edit either, depending on the edit, anything is possible.

What really confused him is the idea that the colorimetric distance between ‘colors’ (device values) in a wider gamut working space is farther apart than in a smaller gamut color space. He mucked up the entire deltaE idea here too; not pertinent, not defined, depends on the color values etc. ALL THINGS being equal, all edits being equal, with an 8-bit per color document, you will very likely end up with banding sooner/greater with a wide gamut working space than a narrower one. So don’t edit in 8-bits per color! In any color gamut working space! No reason you have to and CERTAINLY not a worse outcome than clipping colors you captured and can output with a wider gamut working space when all you have to do, is edit in high bit.

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

Joofa said:

Joofa said:

I used Word with a few find/replace commands to reformat the text file into a form suitable for importing into Excel.

Click to expand...

Ha, if that is really what you do for data 'cleanup' before experimentation, then I guess my comment is really not meant for you.

Click to expand...

What a horror, using Word...

Joofa said:

Joofa said:

Do you really feel the need to discuss this? What is wrong with this forum?

Click to expand...

Do you really feel the need to comment? You are neither OP nor I replied to you. Just keep quiet and work on figuring out your work on Excel.

Click to expand...

Who did you have in mind then? The OP mentioned me explicitly.

Michael Fryd said:

knickerhawk said:

My apology for semi-hijacking this follow-on thread, but I also had a question concerning a post made in the other thread that filled up. Digidog posted a comparison of Granger rainbow/charts constructed in ProPhoto and sRGB (8-bit for both). See here. The respective histograms showed considerable banding in the sRGB version but not in the ProPhoto version. My question to Digidog: How were the Granger rainbows constructed? I can't seem to duplicate the banding in the sRGB versions that I'm creating, so I'd be curious to learn more about the steps used by Digidog. I'm using Photoshop and assume the same was done by Digidog. Thanks, and again my apology for going off-topic in this specific thread. The question didn't seem worthy of a whole separate thread.

Click to expand...

When working with 8-bits per channel, this can happen when converting from a wider gamut colorspace to a smaller gamut colorspace.

Click to expand...

Why would anyone do that if they don’t have to?

Michael Fryd said:

The issue is that within a range of colors, the smaller gamut has more distinct device values.

Click to expand...

Nope.

Michael Fryd said:

In some sense the smaller gamut space has a higher "precision" than the larger gamut space, but the precision of the data has already been reduced to the lower "precision" of the other space. Thus we end up with unused device values in our file.

Click to expand...

Nope.

Michael Fryd said:

This is one of the disadvantages of working in a larger colorspace when your final deliverable will be in a smaller colorspace.

Click to expand...

Wrong.

Michael Fryd said:

You can avoid this problem by working in 16 bits per channel, and converting while still in 16 bits per channel.

Click to expand...

There you go! Problem solved. Work in high bit. New concept for you?

Michael Fryd said:

When it comes to colorspaces, it's important to pay attention to both the theoretical and practical issues.

Click to expand...

And you do that how?

Michael Fryd said:

Sometimes the practical issues will depend on which tools you are using.

Click to expand...

The tools you lack that would provide an analysis of image color gamut it seems.