tech question re eyedroppers?

--I am in the process of going thru Katrin Eismann's great book. I have reset the eyedropper target points to 12 and 243 as she suggests.

I then use the eyedroppers to set the black and white points in the image. The image looks good but when I then check the individual levels readings for R, G and B the sliders have shot past the ends of the histograms. This does not occur when I set the black and white points in the RGB channel manually in levels. When I adjust levels for RGB manually for the same photo and then check the individual level channels, the sliders are still at the end points of each of the 3 histograms. Should not the result in the individual channels be the same regardless of whether I set the black and white points manually or by using the eyedropper method?? Can any of the technically gifted of you explain this to me (in simple terms!)? Thanks.

Bill R

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Bill R

Click to expand...

The eyedropper qpproach puts 12,12,12 and 243,243,243 on the points that you choose visually to be the black and white points. Levals shows you a picture of all the points and the blackest and whitest are at the ends of the diagram. If you happened to choose points that were not the blackest and whitest in the picture you will get the results you described.

You can identify the places in the photo that are the blackest and whitest by holding down "Alt" and "Ctrl" while adjusting the left (black) or right (white) triangles just under the histogram of the levels window.
-bruce

Carl Levinson said:

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Bill R

Click to expand...

The eyedropper qpproach puts 12,12,12 and 243,243,243 on the points
that you choose visually to be the black and white points. Levals
shows you a picture of all the points and the blackest and whitest
are at the ends of the diagram. If you happened to choose points
that were not the blackest and whitest in the picture you will get
the results you described.

Click to expand...

--Thanks Carl and Bruce but I still do not get it. I found the correct black and white points. I also manually changed the RGB scale to 13 and 243 which is where I have set the eyedroppers to. I would think the manual and eye dropper change would therefore cause the same effect. It does on good images but when I try it on a severe red cast image, the 2 methods effect the individual color channelsdifferently. I just do not understand why

Bill R

Bill Richardson said:

--Thanks Carl and Bruce but I still do not get it. I found the
correct black and white points. I also manually changed the RGB
scale to 13 and 243 which is where I have set the eyedroppers to.
I would think the manual and eye dropper change would therefore
cause the same effect. It does on good images but when I try it on
a severe red cast image, the 2 methods effect the individual color
channelsdifferently. I just do not understand why

Bill R

Click to expand...

--I just redid my experimenting on the same red cast photo. The eyedropper method still over corrected the individual color channels when applied to the image with color cast. However, if I corrected the color cast first, then the eye dropper method did not over correct. In fact, the sliders stayed outside the histogram for the individual channels. This seems to indicate that it is better to manually adjust each color in levels for the proper white and black points. OR are there no correct black and white points in the individual color channels but only in the combined RGB channel?? Anybody know???
Bill R

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Bill R

Click to expand...

--

Bill, you might pose your question to Ms. Eismann herself. I bought her book about a year ago, had several questions which I sent to her via email. She got back to me either the same day or next day. Very gracious lady.

Buck

G K said:

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Bill R

Click to expand...

--
Bill, you might pose your question to Ms. Eismann herself. I
bought her book about a year ago, had several questions which I
sent to her via email. She got back to me either the same day or
next day. Very gracious lady.

Buck

Click to expand...

--Great idea Buck. I will finish her book first though. Maybe the answer is in there somewhere. It is a great book--I am learning tons.
Bill R

Bill Richardson said:

G K said:

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Bill R

Click to expand...

--

Click to expand...

Click to expand...

IMHO you have "clipped" too much of each end of the Histogram Try 5 and 250 I think the suggested settings are too drastic The book hints at this being a matter of taste Other PS books suggest a less drastic setting eg Barry Haynes book

Bill Richardson said:

Bill Richardson said:

Bill, you might pose your question to Ms. Eismann herself. I
bought her book about a year ago, had several questions which I
sent to her via email. She got back to me either the same day or
next day. Very gracious lady.

Buck

Click to expand...

--Great idea Buck. I will finish her book first though. Maybe the
answer is in there somewhere. It is a great book--I am learning
tons.
Bill R

Click to expand...

--Robert, I tried the same thing using your adjustments but still got over adjustment in the individual layers. I now think it has something to do with the severe color cast.
Bill R

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Click to expand...

In order to clarify the difference between using sliders in levels and eyedroppers it is best to go to curves which contain the same eyedoppers. Consider the highlight slider in levels. In curves language it simply moves the point at 255,255 over to the left. For example to 255(output),245(input) if you moved the slider to 245. This point simply moves horizontally to the left on the curves graph as you move the highlight slider to the left. Now if one uses the highlight eyedropper the same movement of the 255,255 point ocurrs BUT it moves a different amount. The amount that this point moves to the left is determined by your choice of the highlight point in the image. For example if you chose a highlight point in the image at 223 and adjusted your highlight dropper to go to 245 then the curves point at 255,255 would be moved to the left so that the input value of 223 would have an output value of 245. Try this out using curves so you can see these movements. You can see the curves line corresponding to a slider movement by simply moving the curve point as disecribed above.Unfortunately you cannot verify the change in curves when you move the slider in levels. (The curves graph will remain unchanged because curves only records something done in curves) Since the curves graph is effectively changed differently when one uses a slider or an eydropper it is not surprising that all colors will be modified differently. The curves approach is more versatile of course but the levels approach is easier and more intuitive. I hope this is not too complicated but then Photoshop is not a piece of cake either.

Carl Levinson said:

--Thanks Carl. So if I drag the left point up in curves so the output is 14 and drag the right point down so the value is 243, it is just like setting the levels black and white points. Very interesting. I agree that I will continue to set the points in levels and then switch to curves.

Click to expand...

Bill R

Bill Richardson said:

Bill Richardson said:

--Thanks Carl. So if I drag the left point up in curves so the output is 14 and drag the right point down so the value is 243, it is just like setting the levels black and white points. Very interesting. I agree that I will continue to set the points in levels and then switch to curves.

Click to expand...

Bill R

Click to expand...

Not quite. You drag the left point to the right in curves so that your choice of the shadow point goes to 14. The best way to see this is to use the curves black eyedropper and see what happens. Let's say the point you choose in the image to be the black point reads 20(R),18(G) and16(B) in the info palette. You then click on it in the image with the black eyedropper. It will immediately show up as 14,14,14 in the info palette. If you refer now to the curves representations for the three colors you will see that the bottom left point moved to the right in each case in just such a manner as to make this happen. That is, the input of 20(R) has moved down to output of 14(R) and so on. In each case you are producing curves with an increase in slope which increases the contrast of the image. Incidentally, the middle eyedropper will change the curve in the mid range. Try it out and enjoy. It's a pleasure to behold.

Carl Levinson said:

Carl Levinson said:

Not quite. You drag the left point to the right in curves so that

Click to expand...

your choice of the shadow point goes to 14. The best way to see
this is to use the curves black eyedropper and see what happens.
Let's say the point you choose in the image to be the black point
reads 20(R),18(G) and16(B) in the info palette. You then click on
it in the image with the black eyedropper. It will immediately show
up as 14,14,14 in the info palette. If you refer now to the curves
representations for the three colors you will see that the bottom
left point moved to the right in each case in just such a manner as
to make this happen. That is, the input of 20(R) has moved down to
output of 14(R) and so on. In each case you are producing curves
with an increase in slope which increases the contrast of the
image. Incidentally, the middle eyedropper will change the curve in
the mid range. Try it out and enjoy. It's a pleasure to behold.

Click to expand...

--Carl, Of course!! By switching the white point from 255 to 243 and the black point from 0 to 13 I am increasing contrast which means a steeper curve line. I was changing the output values when I should have been changing the input values and maintaining the full output range of 0 to 255, right? I followed your suggestion and saw it (Ithink) very clearly. It is starting to make sense. Thank you very much. I read a tutorial which suggests using the eyedroppers when there is a definite black and white in the image but using the manual adjustment in levels when there is not. Is that what you do? It would, sometimes, make the levels adjustment unnecessary but it throws in the subjective ability to pick "true" black and whites and I am trying to tke out the guess work. I am glad you find this so interesting.
Bill R

Bill Richardson said:

--I am in the process of going thru Katrin Eismann's great book. I
have reset the eyedropper target points to 12 and 243 as she
suggests.
I then use the eyedroppers to set the black and white points in the
image. The image looks good but when I then check the individual
levels readings for R, G and B the sliders have shot past the ends
of the histograms. This does not occur when I set the black and
white points in the RGB channel manually in levels. When I adjust
levels for RGB manually for the same photo and then check the
individual level channels, the sliders are still at the end points
of each of the 3 histograms. Should not the result in the
individual channels be the same regardless of whether I set the
black and white points manually or by using the eyedropper method??
Can any of the technically gifted of you explain this to me (in
simple terms!)? Thanks.

Click to expand...

There is a subtle point here. Check out page 76 in Eismann where she does not arbitrarily decide the white point should go to 245 or so. She looks for the best candidate for the white point in the image itself and forces the smaller color numbers in this point to move up to the brighest one thus producing a gray scale point when R=G=B. This can be done in either curves or as she does in levels. In this approach one does not use the eydroppers at all but lets the image itself determine the strength of the three points.(Highlight, mid and Shadow) In this approach the contrast is not affected as much which may be good or bad depending on artistic considerations. So ulitimately there is no "by the numbers" solution which gives the best result only various approaches which differ and are determined by taste.

Carl Levinson said:

Carl Levinson said:

There is a subtle point here. Check out page 76 in Eismann where

Click to expand...

she does not arbitrarily decide the white point should go to 245 or
so. She looks for the best candidate for the white point in the
image itself and forces the smaller color numbers in this point to
move up to the brighest one thus producing a gray scale point when
R=G=B. This can be done in either curves or as she does in levels.
In this approach one does not use the eydroppers at all but lets
the image itself determine the strength of the three
points.(Highlight, mid and Shadow) In this approach the contrast is
not affected as much which may be good or bad depending on artistic
considerations. So ulitimately there is no "by the numbers"
solution which gives the best result only various approaches which
differ and are determined by taste.

Click to expand...

--Carl, Thanks. I have gone back and reread p 76. I had been focusing only on the neutral gray but I will look at all 3 now (black, gray and white) Your previous response inspired me to do a lot of thinking and experimenting. I have been using curves and have been able to visualize most of the effects of moving points around before doing it. I therefore think I am starting to grasp curves with one big exception--hues. This is the way I understand it and would much appreciate you telling me if I am correct: I think I am right in concluding that desaturation is the result of reducing the tonal range of the OUTput range thereby forcing the INput tonal range into the gray midtones of the OUTput range. Therefore, if I flatten the curve line to horizontal, everything is desaturated and turns gray. If I steepen the curves line to vertical I am forcing the most intense color tones (the peak of the Histogram) into the full OUTput range thereby causing an increase in color intensity across the entire OUTput range. So I am thinking that hue is color intensity WITHIN THE MIDTONE GRAYS. Right? Thanks for your patience!!!
Bill R

Bill Richardson said:

Carl Levinson said:

Bill Richardson said:

There is a subtle point here. Check out page 76 in Eismann where

Click to expand...

she does not arbitrarily decide the white point should go to 245 or
so. She looks for the best candidate for the white point in the
image itself and forces the smaller color numbers in this point to
move up to the brighest one thus producing a gray scale point when
R=G=B. This can be done in either curves or as she does in levels.
In this approach one does not use the eydroppers at all but lets
the image itself determine the strength of the three
points.(Highlight, mid and Shadow) In this approach the contrast is
not affected as much which may be good or bad depending on artistic
considerations. So ulitimately there is no "by the numbers"
solution which gives the best result only various approaches which
differ and are determined by taste.

Click to expand...

--Carl, Thanks. I have gone back and reread p 76. I had been
focusing only on the neutral gray but I will look at all 3 now
(black, gray and white) Your previous response inspired me to do a
lot of thinking and experimenting. I have been using curves and
have been able to visualize most of the effects of moving points
around before doing it. I therefore think I am starting to grasp
curves with one big exception--hues. This is the way I understand
it and would much appreciate you telling me if I am correct: I
think I am right in concluding that desaturation is the result of
reducing the tonal range of the OUTput range thereby forcing the
INput tonal range into the gray midtones of the OUTput range.
Therefore, if I flatten the curve line to horizontal, everything is
desaturated and turns gray. If I steepen the curves line to
vertical I am forcing the most intense color tones (the peak of the
Histogram) into the full OUTput range thereby causing an increase
in color intensity across the entire OUTput range. So I am thinking
that hue is color intensity WITHIN THE MIDTONE GRAYS. Right?
Thanks for your patience!!!
Bill R

Click to expand...

-- Bill There is much to be said about Hue and Saturation. I wil mention the key concepts and then leave it to personal E-Mails to follow up with the details since I do not want to clog up the Forum with stuff which may not be interesting to others.

Let R,G,B stand for the three color definitons in DESCENDING order. That is R is the largest number and so on to Blue the smallest. Now the color picker will give you values of H,S,Br (I use Br for Brightness)for any values of R,G,B,. So by playing around with the color picker one can uncover what the color picker means by H,S,Br in terms of R,G,B. I have worked this out and it leads to the following picture. - - Divide the original pixel into the sum of two pixels, a Saturated pixel (R-B,G-B,0) and an unsaturated pixel (B,B,B). The Saturated pixel determines the Hue and the Saturation is measured by the Ratio of the Saturated pixel's brightness to the total brightness of the original pixel.The Brightness of any pixel is perportional to the value of the largest color. This differs from the Luminosity which is a much better measure of the observed brightness of a pixel since the Luminosity takes into account all three numbers in a pixel. I have worked this out also and find that Luminosity weights the colors Red, Green and Blue differently which must correspond to actuall tests on human subjects. I have not thought much about your suggestion of relating all this to curves. Perhaps we can continue by E-Mail unless there is some expressed interest by others in this.
Carl

Carl Levinson said:

. I have
not thought much about your suggestion of relating all this to
curves. Perhaps we can continue by E-Mail unless there is some
expressed interest by others in this.
Carl

Click to expand...

--Thanks Carl. I will be studying yourcomments.
Bill R