Ankor Thom--Let's call it "Framed Statues"

[mikemeister_admin]

These images are in progress.  I have three neutrals, one being in the middle and one each at higher and lower points.  I closed the image before I tested trying to move all three, but are there advantages to three neutrals?

The second image is one in which I manually set a point fairly high in the curve and manipulated the numbers slightly to make them all the same value.  I then set a neutral as close to that point as possible (not shown)  OH!  The difference between manually setting a hue clock and making the numbers equal and setting a neutral is that setting a neutral is global, whereas manipulating the numbers on a hue clock is local.  Well--Duh!  Another insight!  Man I hope no one else reads this, as I'm sure they figured this out long ago.  Better late than never!  So, please see if this is correct:  in setting a neutral, one tries to find a spot in which the numbers are somewhat equal, and the result is a global change that shifts all the colors in the image?  But shifts them all more or less in the same proportion as the numbers in the neutral, or in accordance to the set neutral?  Or is it a complex algorithm that would be way out there in math la-la land?

On a third try, I set two neutrals fairly far apart--not really neutrals.  However when I adjust, say, the red, at either the high point or the lower one, the blue curve moves MUCH more than the red or green, sometimes contorting the curve way out of line with the other two channels, giving a blue cast.  Is this because I put the points too far apart?  I still wouldn't think that would make the blue curve more sensitive.  Does it matter which channel you chose to manipulate all three channels up and down?  This is not a final image--just ones I'm posting with questions.  I can make it look better, but I wanted to seek answers to the issues posed by the points that I set.

It really bothers me that the highlights in the left doorway are so blown out.  Is there a burn tool in Curvemeister?!

Thanks so much!

[ggroess]

don't just think of them as Gray and mid tone...A neutral can be black or white as well..

Greg

[mikemeister_admin]

You could try this as your first layer.Ten points is a bit more than I would usually use and I wouldn't do it if there was a lot of color information between 192 and 255,colors would be affected.GregM

[mikemeister_admin]

Hi Greg,

Yes, thanks for the reminder (again) that a neutral can be at any point in the scale.  In this terminology, neutral means without color cast, right?  If this is the case, why did you tell us to look for something that we were pretty sure was gray?

Sorry I had such a long list of questions from above but, are there advantages to setting three floating neutrals?  Without a locking point in the middle of the curve, moving either of the two floating neutrals up or down resulted in a really out of proportion blue channel.  I don't really see a point to floating neutrals if it is going to skew one of the channels more than the others.

Thanks!  Rhonda


I know you probably just didn't get to it, but also from my earlier post:
In setting a neutral, one tries to find a spot in which the numbers are somewhat equal, and the result is a global change that shifts all the colors in the image?  But shifts them all more or less in the same proportion as the numbers in the neutral, or in accordance to the set neutral?  Or is it a complex algorithm that would be way out there in math la-la land?

[mikemeister_admin]

To gremurphogo:  you said:  You could try this as your first layer.Ten points is a bit more than I would usually use and I wouldn't do it if there was a lot of color information between 192 and 255,colors would be affected.GregM

Sorry, I don't have a deep enough understanding of colors to know if I have a lot of color information between 192 and 255.  Do you mean by looking at the histogram of the channels?  Why would the number of points affect the colors that you were trying to improve by setting the points in the first place?  If the answer to these questions is complex, better for you not to try!  :)

[ggroess]

These images are in progress.  I have three neutrals, one being in the middle and one each at higher and lower points.  I closed the image before I tested trying to move all three, but are there advantages to three neutrals?

Not really, but if you do a by the numbers correction for S/H/Mid-tone you are in effect doing a "3 Neutrals" correction it's really just another way of looking a the same problem...Why do it...some people find it faster...Trouble is that images with a true mid-tone Gray are sometimes hard to find.  Also if you have known neutrals you can adjust the contrast fast using the channels rather than the master channel..no color cast introduced.

The second image is one in which I manually set a point fairly high in the curve and manipulated the numbers slightly to make them all the same value.  I then set a neutral as close to that point as possible (not shown)  OH!  The difference between manually setting a hue clock and making the numbers equal and setting a neutral is that setting a neutral is global, whereas manipulating the numbers on a hue clock is local.  Well--Duh!  Another insight!  Man I hope no one else reads this, as I'm sure they figured this out long ago.  Better late than never!  So, please see if this is correct:  in setting a neutral, one tries to find a spot in which the numbers are somewhat equal, and the result is a global change that shifts all the colors in the image?  But shifts them all more or less in the same proportion as the numbers in the neutral, or in accordance to the set neutral?  Or is it a complex algorithm that would be way out there in math la-la land?

If you set a neutral on a point CM calculates the average of the original points and places the neutral there.  I just tested this on an image with R=134 G=137 and B = 147.  CM set the neutral at 140.    The difference really is that we sometimes "fudge" the numbers a bit to make the correction local...you are correct...  CM setting a neutral will average the original sample point.

On a third try, I set two neutrals fairly far apart--not really neutrals.  However when I adjust, say, the red, at either the high point or the lower one, the blue curve moves MUCH more than the red or green, sometimes contorting the curve way out of line with the other two channels, giving a blue cast.  Is this because I put the points too far apart?  I still wouldn't think that would make the blue curve more sensitive.  Does it matter which channel you chose to manipulate all three channels up and down?  This is not a final image--just ones I'm posting with questions.  I can make it look better, but I wanted to seek answers to the issues posed by the points that I set.

I was able to have the pins become disconnected like you did...Please watch to make sure that all three curves are moving together for this to work.  I think it is possible to unlink the points...I was not able to make it happen a second time... yet... I did move the red curve independent of the others.  This is most likely what is happening to you.

It really bothers me that the highlights in the left doorway are so blown out.  Is there a burn tool in Curvemeister?!

 

No but you can try other things like masking to see if you can make this right....
Further...

To gremurphogo:  you said:  You could try this as your first layer.Ten points is a bit more than I would usually use and I wouldn't do it if there was a lot of color information between 192 and 255,colors would be affected.GregM

Sorry, I don't have a deep enough understanding of colors to know if I have a lot of color information between 192 and 255.  Do you mean by looking at the histogram of the channels?  Why would the number of points affect the colors that you were trying to improve by setting the points in the first place?  If the answer to these questions is complex, better for you not to try!  :)

Rhonda,
You know enough to know the highlights are blown out a bit...If you click on the highlight threshold in RGB you will see that there is color information in the frame of the door.  Greg M is making a fair observation and stating his preference.  I'm not going to put words in his mouth;  I'd say try it both ways and see which you prefer.  I believe he is saying changing the highlight by that much might make it worse or make it stick out...
Greg

[mikemeister_admin]

I used a few points close together to stop the diagonal from assuming an S shape when I lowered the top right and affecting the main body of the image.As the part of the diagonal you're bending  becomes more horizontal the more dingy  the highlight area becomes.The range of correction very small,but works on this image to tame that wall a bit.The location of the pins was to con tain the effect to the  area approaching 255,255,255 white.

In the histogram below the new blob of pixels suggests an area of an image with detail and color and a move like I described above  could affect a larger,different area of the image.I would rather make brightness correction to this image in Lab or HSB then move to RGB for the color.GregM