LAB changes certainly brought out the colour well.
A couple of clarifications.
Am I correct that to remove color cast you set the neutral point?
The set neutral point is where the color is as close as possible to 50,0,0 (L,a,b)? Is this equivalent to 18% grey? So if I have a color card in the shot (like example 1) I simply find the grey that has a 50,0,0 color setting and set this as neutral?
The set highlight point is as close as possible to 100,0,0?
The set shadow point is 0,0,0?
Are these the ideal settings we are looking for and can check in the hue clock as we move the point over the image.
Benny
Benny Class 1 Session 2 Example 4
Setting the neutral tells the CM tool where A and B zero are yes. If they are not actually zero in the image then CM sets them to be zero this is why the understanding of neutrals is so important. Pure white, pure black, and greys that are neutral are good candidates for a neutral point.
In LAB the neutral is really when the a and b are both zero, yes. The L channel could be more or less than 50 but the A and B at zero puts the color at the center of both of the grids.
I'm uncertain about the 18% grey question that would be more Mike's area. I think it's closer to 70.
If you are looking to set up controls, using the grey square is important yes but so is a monitor calibration profile and a printer profile. In the past the discussions have been very involved regarding monitor calibraton and tools to do this. Even if you shoot a good grey card, and match the color on the monitor, if the monitor settings are wrong your file will have a color shift in it for printing. That is where tools like the Spyder color checker come in.
I would say that for a given exposure the highlight and shadow really depend on the accuracy of the exposure. If the histogram does not have a 100, 0, 0 value then you have to adjust the image to achieve one. you are resetting the scale so to speak and making a pixel value of say 92,0,0 show as 100,0,0 and adjusting all the pixels along with it.
The toughest thing to teach in this class is the subjective part of the process. When you say "Ideal" I start to think that you might get hung up in having to have the exact values...many images do not in fact have the 100 or the 0 "L" values. The entire process is really about compromise. You can get hung up in chasing your image around in circles to get it back where you want it.
Personally, I work on an image until I feel like I'm getting diminishing returns. When I make a change to an acceptable image and I feel like it's worse...then it is time to stop. I also work constantly on not over correcting the image. It is very easy to take a acceptable image and make it unbelieveable with the click of the mouse.
I use the hue clock to check what my eyes are telling me. If the neutral I set just does not look right I check it with the hue clock. I also use it to help me when I look at images from the web to see if the browser is interpreting th image differently. That is why you will see me ask sometimes if the image looks better to the person that posted it. I have added images to the site and had them look nothing like they do in photoshop.
Sorry for the long winded answer but those were excellent questions.
BTW your moth looks pretty good...
Greg
In LAB the neutral is really when the a and b are both zero, yes. The L channel could be more or less than 50 but the A and B at zero puts the color at the center of both of the grids.
I'm uncertain about the 18% grey question that would be more Mike's area. I think it's closer to 70.
If you are looking to set up controls, using the grey square is important yes but so is a monitor calibration profile and a printer profile. In the past the discussions have been very involved regarding monitor calibraton and tools to do this. Even if you shoot a good grey card, and match the color on the monitor, if the monitor settings are wrong your file will have a color shift in it for printing. That is where tools like the Spyder color checker come in.
I would say that for a given exposure the highlight and shadow really depend on the accuracy of the exposure. If the histogram does not have a 100, 0, 0 value then you have to adjust the image to achieve one. you are resetting the scale so to speak and making a pixel value of say 92,0,0 show as 100,0,0 and adjusting all the pixels along with it.
The toughest thing to teach in this class is the subjective part of the process. When you say "Ideal" I start to think that you might get hung up in having to have the exact values...many images do not in fact have the 100 or the 0 "L" values. The entire process is really about compromise. You can get hung up in chasing your image around in circles to get it back where you want it.
Personally, I work on an image until I feel like I'm getting diminishing returns. When I make a change to an acceptable image and I feel like it's worse...then it is time to stop. I also work constantly on not over correcting the image. It is very easy to take a acceptable image and make it unbelieveable with the click of the mouse.
I use the hue clock to check what my eyes are telling me. If the neutral I set just does not look right I check it with the hue clock. I also use it to help me when I look at images from the web to see if the browser is interpreting th image differently. That is why you will see me ask sometimes if the image looks better to the person that posted it. I have added images to the site and had them look nothing like they do in photoshop.
Sorry for the long winded answer but those were excellent questions.
BTW your moth looks pretty good...
Greg
Greg makes some good points.
I would add that setting the neutral at RGB(128,128,128) is not the goal at all. What you want is the other way around. Look for something in the image that that you know, from your own experience, should be gray, but is not. That's the color cast you want to remove.
In this case I give you the clue that the round spots on the wings should be neutral. When you create a neutral at such a location, it will remove the color cast from that area, and remove the color cast from the rest of the image. It's not important that the gray be exactly middle gray, but it's better if it's somewhere in the general middle range of the image, since colors are more intense.
Lab only allows one neutral, but in RGB mode you can set two or even three neutrals, provided they are of different brightnesses. Watch the curve to see if the neutrals are crowding one another and causing the curve to buckle. That's your clue that the neutrals are too close to one another.
Re 18 percent gray, the short anwer is that it works out to about RGB(118,118,118).
I would add that setting the neutral at RGB(128,128,128) is not the goal at all. What you want is the other way around. Look for something in the image that that you know, from your own experience, should be gray, but is not. That's the color cast you want to remove.
In this case I give you the clue that the round spots on the wings should be neutral. When you create a neutral at such a location, it will remove the color cast from that area, and remove the color cast from the rest of the image. It's not important that the gray be exactly middle gray, but it's better if it's somewhere in the general middle range of the image, since colors are more intense.
Lab only allows one neutral, but in RGB mode you can set two or even three neutrals, provided they are of different brightnesses. Watch the curve to see if the neutrals are crowding one another and causing the curve to buckle. That's your clue that the neutrals are too close to one another.
Re 18 percent gray, the short anwer is that it works out to about RGB(118,118,118).
Return to “May 2007 Curvemeister 101”
Who is online
Users browsing this forum: No registered users and 24 guests