login info join!
Forums > Digital Art and Retouching > measuring the relative amount of blue in a sample Search   Reply
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


how do you measure the relative amount of blue in a sample group of pixels? I want to compare the amount of blue in one spot/patch to another patch in the same image.

Thanks,

Jeff
Feb 20 13 11:49 am  Link  Quote 
Retoucher
Krunoslav Stifter
Posts: 3,847
Santa Cruz, California, US


How big the number of pixels? If it's small enough the Photoshop's eyedroper tool can be set up to display average sample point ranging from 1 pixel to grid of 101 by 101 px average.

If it's like a whole image than not sure, but if you say; "how do you measure the relative amount of blue in a sample group of pixels?" Maybe you can try to select the part of the image that you want and use Filter > Blur > Average. And than read out what the color is and based on the RGB numbers in the info panel you should be able to see what is the; "relative amount of blue in a sample group of pixels" ...but this is only if I'm understanding you correctly and that is in fact what you want to do.
Feb 20 13 12:04 pm  Link  Quote 
Photographer
BTHPhoto
Posts: 6,771
Fairbanks, Alaska, US


In photoshop, your histogram can be set to display just the blue channel, or you can choose all channels, or you can choose "colors" to get a single graph with overlaid histograms of red, green, and blue channels.  If you select an area, it will display the histogram for just the pixels in that area.
Feb 20 13 12:12 pm  Link  Quote 
Photographer
Motordrive Photography
Posts: 2,480
Lodi, California, US


Krunoslav-Stifter wrote:
info panel you should be able to see what is the; "relative amount of blue in a sample group of pixels" ...but this is only if I'm understanding you correctly and that is in fact what you want to do.

+1
info palette can give you RGB, CMYK or HSB numbers

Feb 20 13 12:47 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


My retoucher is pointing out some patches of an image that appear to have a color cast. I wanted to use diff raw conversions and compare the amount of blue in these patches relative to other areas in the same image where we don't "see" the blue.

patch #1: RGB is 160, 125, 98
patch #2: RGB is 227, 190, 163

Which has more blue & why? Don't say #2 because of the 163 value. These are absolute values, meaning it is higher because the exposure/brightness is higher. That's what I mean by relative.

thanks,

Jeff
Feb 20 13 01:05 pm  Link  Quote 
Photographer
Motordrive Photography
Posts: 2,480
Lodi, California, US


Jeffery Williams wrote:
patch #1: RGB is 160, 125, 98
patch #2: RGB is 227, 190, 163

Jeff

they are almost identical color (1 degree difference)
they are vastly different in saturation and brightness

usually values are stated as R-G-B, instead of the other way around

Feb 20 13 01:25 pm  Link  Quote 
Photographer
David Kirk
Posts: 4,430
Ottawa, Ontario, Canada


I'm no colour expert, but have you tried to use HSL instead of RGB and then compare the hue only so that the saturation and brightness are not a factor?
Feb 20 13 01:32 pm  Link  Quote 
Photographer
Motordrive Photography
Posts: 2,480
Lodi, California, US


smile yes, I meant to say hue, but the point was the same
Feb 20 13 01:40 pm  Link  Quote 
Photographer
Ruben Vasquez
Posts: 3,104
Puyallup, Washington, US


Jeffery Williams wrote:
My retoucher is pointing out some patches of an image that appear to have a color cast. I wanted to use diff raw conversions and compare the amount of blue in these patches relative to other areas in the same image where we don't "see" the blue.

patch #1: RGB is 160, 125, 98
patch #2: RGB is 227, 190, 163

Which has more blue & why? Don't say #2 because of the 163 value. These are absolute values, meaning it is higher because the exposure/brightness is higher. That's what I mean by relative.

thanks,

Jeff

The red channel is the dominant color and the blue channel is contributing the least. Based on those numbers alone and without knowing which rgb color space you're in, I would guess there is no blue but would think that color would be like a reddish orange.

I would recommend setting up your info panel to read lab is that is the most straight forward way of comparing color. To measure blue, simply look at how negative the B channel is. 0 will be completely neutral (i.e. black, white or grey), while something like -5 has a very slight blueish cast and something like -30 will be very blue and so on. Positive numbers in the B channel on the other hand will be yellow. Positive numbers in the A channel are magenta and negative numbers are green.   

You can use rgb but if you're going by the numbers, you have to take into account how the other numbers will affect that patch of pixels as well as what rgb space you're in (i.e srgb vs argb). Likewise you can use hsb but you have to know what range of hue values correlate to blue before you can look at saturation. Lab just tells you how much of what color you have. Very simple and easy to read once you understand how it's set-up.

Feb 20 13 01:46 pm  Link  Quote 
Photographer
TMA Photo and Retouch
Posts: 712
New York, New York, US


If you use the "Color Sampler Tool"...you will be able to place down up to 4 color sample points and view all of their color numbers.

You will also notice in the tool option bar... that there is a chooser / pull down.  It says Point Sample, 3 pixels, 5 pixels, 11 pixels, 31 pixels etc. These various options give you an average of all the color numbers inside that larger square area.  Great for averaging colors in a specific defined area.

The color points will all show up in the info pallet as point #1, #2 etc...and you can also change it to read the RGB, or LAB, or CYMK numbers too. 

Note: In the Point Sample mode you are sampling only 1 single pixel.  Its easy to sample just 3 pixels to the left... and get really different numbers...even though you think the colors should be the same.  Thats why the color average feature becomes so helpful.

If you are sharing info between you and your retoucher...just place your points...save your image as a psd... and then send it to your retoucher.  He will now have the exact same points as you chose... so there can be no disagreement in sample points and numbers.

You are right...if you increase just 1 number...it will affect to a small degree the underlying luminance value as well as the hue.

Cheers,

Ray
Feb 20 13 01:57 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Ruben Vasquez wrote:

The red channel is the dominant color and the blue channel is contributing the least. Based on those numbers alone and without knowing which rgb color space you're in, I would guess there is no blue but would think that color would be like a reddish orange.

I would recommend setting up your info panel to read lab is that is the most straight forward way of comparing color. To measure blue, simply look at how negative the B channel is. 0 will be completely neutral (i.e. black, white or grey), while something like -5 has a very slight blueish cast and something like -30 will be very blue and so on. Positive numbers in the B channel on the other hand will be yellow. Positive numbers in the A channel are magenta and negative numbers are green.   

You can use rgb but if you're going by the numbers, you have to take into account how the other numbers will affect that patch of pixels as well as what rgb space you're in (i.e srgb vs argb). Likewise you can use hsb but you have to know what range of hue values correlate to blue before you can look at saturation. Lab just tells you how much of what color you have. Very simple and easy to read once you understand how it's set-up.

The b channel is showing 26-29 in the questionable areas. The other areas are about the same. So, we can't count on eye's is the lesson I am getting. Working in ARGB.

Feb 20 13 02:01 pm  Link  Quote 
Photographer
TMA Photo and Retouch
Posts: 712
New York, New York, US


Duplicate - Erased
Feb 20 13 02:06 pm  Link  Quote 
Retoucher
Peano
Posts: 4,106
Lynchburg, Virginia, US


Jeffery Williams wrote:
My retoucher is pointing out some patches of an image that appear to have a color cast.

If a (possible) color cast is the problem, I think you're going at it the wrong way. You need to find what should be true whites, blacks, or neutral grays in between, and check those RGB values.

EDIT: If those are skin tones that you posted the values for, then they look to me like they have too much yellow (too little blue) in them.

Feb 20 13 02:24 pm  Link  Quote 
Photographer
Ruben Vasquez
Posts: 3,104
Puyallup, Washington, US


Jeffery Williams wrote:
The b channel is showing 26-29 in the questionable areas. The other areas are about the same. So, we can't count on eye's is the lesson I am getting. Working in ARGB.

That's right. Regardless of if the monitor(s) is calibrated and you have the proper viewing conditions, our eyes are easily deceived and can't be trusted to make objective, color critical decisions which is where the numbers are particularly useful.

Our eyes are awfully good figuring out what looks good or not though. Those areas are pretty well saturated with yellow, the only question now, is that to your liking?

Feb 20 13 02:34 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Peano wrote:

If a (possible) color cast is the problem, I think you're going at it the wrong way. You need to find what should be true whites, blacks, or neutral grays in between, and check those RGB values.

EDIT: If those are skin tones that you posted the values for, then they look to me like they have too much yellow (too little blue) in them.

The issue is that patches in the image, not the entire image, "appear" to have a blue tint. My first reaction was to prove this true or false. I now believe our eyes are deceiving us. An illusion if you will. Yes, skin tones.

Feb 20 13 02:42 pm  Link  Quote 
Retoucher
Krunoslav Stifter
Posts: 3,847
Santa Cruz, California, US


Jeffery Williams wrote:
I now believe our eyes are deceiving us. An illusion if you will. Yes, skin tones.

Not sure if you can download and watch this, but I think it might confirm your suspicions? smile

http://dl.dropbox.com/u/28508998/Jeff/W … yesMAC.zip

Feb 20 13 02:51 pm  Link  Quote 
Photographer
Atelier Hereau
Posts: 69
Stoughton, Massachusetts, US


Jeffery Williams wrote:
My retoucher is pointing out some patches of an image that appear to have a color cast. I wanted to use diff raw conversions and compare the amount of blue in these patches relative to other areas in the same image where we don't "see" the blue.

patch #1: RGB is 160, 125, 98
patch #2: RGB is 227, 190, 163

Which has more blue & why? Don't say #2 because of the 163 value. These are absolute values, meaning it is higher because the exposure/brightness is higher. That's what I mean by relative.

thanks,

Jeff

By simple % calculation, the % blue in Patch #1 is 25.59, the % blue in Patch #2 is 28.10.  #2 has 9.8% more blue than #1.

Is that what you you want to know?

Feb 20 13 03:41 pm  Link  Quote 
Retoucher
Peano
Posts: 4,106
Lynchburg, Virginia, US


Perception is reality. Don't get hung up on numbers.
Feb 20 13 04:01 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Rich Images wrote:

By simple % calculation, the % blue in Patch #1 is 25.59, the % blue in Patch #2 is 28.10.  #2 has 9.8% more blue than #1.

Is that what you you want to know?

That's what I did but didn't know if was a valid measure.

Feb 20 13 04:17 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Peano wrote:
Perception is reality. Don't get hung up on numbers.

Oh great, now I have to use my eyes? smile

Feb 20 13 04:18 pm  Link  Quote 
Retoucher
Peano
Posts: 4,106
Lynchburg, Virginia, US


Jeffery Williams wrote:
Oh great, now I have to use my eyes? smile

Mother Nature has a way of getting the last word . . .

http://img441.imageshack.us/img441/5663/braini.jpg

Feb 20 13 04:31 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Krunoslav-Stifter wrote:

Not sure if you can download and watch this, but I think it might confirm your suspicions? smile

http://dl.dropbox.com/u/28508998/Jeff/W … yesMAC.zip

downloaded fine, ~20 mb file but got an error when running, "server something"

Feb 20 13 04:36 pm  Link  Quote 
Retoucher
Krunoslav Stifter
Posts: 3,847
Santa Cruz, California, US


Jeffery Williams wrote:
downloaded fine, ~20 mb file but got an error when running, "server something"

Yeah, that's weird. Some pople have issues others don't, not sure why. Will upload again. Wait.

OK, Try now.

http://dl.dropbox.com/u/28508998/Jeff/W … r-Eyes.zip

Feb 20 13 04:38 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Krunoslav-Stifter wrote:

Yeah, that's weird. Some pople have issues others don't, not sure why. Will upload again. Wait.

OK, Try now.

http://dl.dropbox.com/u/28508998/Jeff/W … r-Eyes.zip

same issue sad

Feb 20 13 06:09 pm  Link  Quote 
Retoucher
Krunoslav Stifter
Posts: 3,847
Santa Cruz, California, US


Jeffery Williams wrote:

same issue sad

I have no idea why that is, only video that some people can play without a problem, including me and for some without any explanation it's a problem. Hmmm. I have exported, encoded the entire video again and uploaded it. Hope it works.

http://dl.dropbox.com/u/28508998/Jeff/W … r-Eyes.zip

Feb 20 13 06:51 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Here's an update and the image

http://jefferywilliamsphotography.com/shoots/cyanPoints.jpg

Notice the blue tint in the chest area and the C % numbers. I am not sure what this is really telling us. Should the C % be pretty constant throughout the skin? A quick test on another image say's no. Again, the issue is not that the skin tone is bad or not color/WB corrected. Instead, it is that parts of the chest area appear to have a [more] blue tint than the rest of the skin. Are the C % numbers proof of this? If yes, then is this something to worry about? What causes it?

Image taken with D800, raw conversion with capture nx2, argb color spare, ab lights, wb=flash, wb correction using gray card. No ambient light.
Feb 21 13 04:43 pm  Link  Quote 
Photographer
Ruben Vasquez
Posts: 3,104
Puyallup, Washington, US


Jeffery Williams wrote:
Here's an update and the image

Notice the blue tint in the chest area and the C % numbers. I am not sure what this is really telling us. Should the C % be pretty constant throughout the skin? A quick test on another image say's no. Again, the issue is not that the skin tone is bad or not color/WB corrected. Instead, it is that parts of the chest area appear to have a [more] blue tint than the rest of the skin. Are the C % numbers proof of this? If yes, then is this something to worry about? What causes it?

Image taken with D800, raw conversion with capture nx2, argb color spare, ab lights, wb=flash, wb correction using gray card. No ambient light.

I don't see a blue tint no. What I see is that the chest is less saturated especially in comparison to her arm so the numbers don't surprise me. Red is the opposite of cyan so the more red you have (like her arm), the lower the cyan percentile will be where as her chest has less red in it so the percent cyan is higher. There are natural variations in skin tone (both in luminance and in color), throughout the body and that of course, is going to vary from person to person.

Feb 21 13 06:59 pm  Link  Quote 
Photographer
Jeffery Williams
Posts: 311
Tampa, Florida, US


Ruben Vasquez wrote:

I don't see a blue tint no. What I see is that the chest is less saturated especially in comparison to her arm so the numbers don't surprise me. Red is the opposite of cyan so the more red you have (like her arm), the lower the cyan percentile will be where as her chest has less red in it so the percent cyan is higher. There are natural variations in skin tone (both in luminance and in color), throughout the body and that of course, is going to vary from person to person.

Thank you Ruben

Feb 21 13 07:03 pm  Link  Quote 
Retoucher
Peano
Posts: 4,106
Lynchburg, Virginia, US


Jeffery Williams wrote:
Notice the blue tint in the chest area and the C % numbers. I am not sure what this is really telling us. Should the C % be pretty constant throughout the skin?

It isn't telling you anything. You're over-analyzing. Open a selective color adjustment layer and set it like this:

http://img685.imageshack.us/img685/3138/selcol.jpg

Invert the mask and paint white with a soft brush, 10% brush opacity, over the parts that look a tad blue to you.

Feb 21 13 07:27 pm  Link  Quote 
  Search   Reply



main | browse | casting/travel | forums | shout box | help | advertising | contests | share | join the mayhem

more modelmayhem on: | | | edu

©2006-2014 ModelMayhem.com. All Rights Reserved.
MODEL MAYHEM is a registered trademark.
Toggle Worksafe Mode: Off | On
Terms | Privacy | Careers