• Welcome to BirdForum, the internet's largest birding community with thousands of members from all over the world. The forums are dedicated to wild birds, birding, binoculars and equipment and all that goes with it.

    Please register for an account to take part in the discussions in the forum, post your pictures in the gallery and more.
ZEISS DTI thermal imaging cameras. For more discoveries at night, and during the day.

A question of Levels (1 Viewer)

Roy C

Occasional bird snapper
I usually adjust the levels in CS4 by using the RGB channel - is there any merit in adjusting the Red, Green and Blue Channels individually instead? I have played around with adjusting the individual channels but cannot make my mind up if it is beneficial.
While on the subject I notice that it is almost always the Red channel that is clipped for the whites is there any other part of processing that would benefit from using channels?
 
Altering each individual channel will adjust the colour balance. It sometimes works well but can also look awful if overdone!

The middle slider will intensify the colour selected by moving left and reduce the colour by moving right. So if you've got, say, a slightly 'cold' image that you don't want to re-do all the way back from a Raw, you can move the slider over to the right in the blue channel to warm it up a bit.

In theory, moving the outer sliders in each channel up to 'nearly' touching the start of the 'wedge' of the histogram 'should' give a correct colour balance - but I'm not convinced of this from some experiments I've tried:eek!:
 
Roy,

Depends on what you are trying to do with levels either color balance or increase the tonal range of an image.
I prefer curves to levels, since it gives me more control over both and it is the first part of my workflow.
If my image is a little too warm, I select the red channel in curves and adjust the curve to reduce the reds etc.
Generally, I use curves to set my black points and white points, which once done, takes care of balancing the image since the midtones fall into place as a result. Here’s my workflow, the steps of which may be common knowledge, but hopefully helpful. It’s more simple than it seems once you have it set up.

In curves, click on the black eyedropper in the dialog box and when the color picker come up, set the values in the RGB checkboxes to 10, 10, 10. Close and click Click on the white eyedropper and set the same values to 242,242,242. Don’t touch the grey (midtone) eyedropper.
The reason for doing this is - think of a stop-wedge going from black to grey to white, the values of black starting at 0 and ending with white that has a value of =255. 0 equals blacks with no detail and 255 is “paper” white eg. an area that has no details, referred to as clipped or “burnt-out”.
Setting the black values to 10 and whites to 242 (or245) basically retains details in the whites and as much detail in the shadows (I found anything from a value between 1-10 in the blacks is not discernible in print, so no point setting it below 10). Some will argue, but I set mine this way and prints in magazines etc all come out great.
Close curves and when it asks if you want to save the new targets as defaults, click “yes”. This basically telling PS to map any true blacks in the image to a value of 10 and whites to 242.

Now, let’s say I want to quickly color balance a picture of an Ivory Gull. Open the info palette Window>Info, so it’s visible, from the tool menu, click on the eyedropper tool and as you move it over what you consider the darkest part of the image, or a true black, keep an eye on the readings in the info palette. Once you find the darkest spot (values reading close to 0) click to place a sampler at that spot. Odds are that all the RGB reading in the info palette will be close but not equal.
At this point, open up curves, click on the black eyedropper tool and carefully click on the sampler you placed on your image. It will make those RGB values 10,10,10. Since each channel has the same value, it basically renders the image neutral, thus color balanced.


Now, use the eyedropper tool again and move it over the whitest areas, or brightest areas and look at the info palette (values of 242-250). Once found, click a sampler in place. Open up curves, click on the white eyedropper and click on the sampler you just placed. All the RGB values will change to reflect 242,242,242. Since each channel has the same value, it basically renders the image neutral, thus color balanced.

NOTE1 – this works best on nicely lit images –images taken at sunrise or sunset will not benefit since this workflow will eliminate the nice warm colors that are desirable. Also, it is important to make sure you click on the darkest/whites parts. Clicking on a dark brown for example, or a dull white will not produce pleasing results.

NOTE2 – to find the darkest or whites parts of an image, open up Levels and start to move the black (shadow) slider inwards - hold the ALT (PC) down as you do this and the screen goes white. Move the slider until you start to see the first black/colored pixels appear. These are the areas of the image that are the darkest, or first areas to be ‘clipped’. Do the same with the white slider and the screen goes black. Move the slider until you start to see the first colored pixels appear. These are the areas of the image that are the whitest, or first areas to be ‘clipped’. Remember this area in the image so that you can quickly home in on that area instead of hunting for the brightest, darkest area. Then follow my instructions.

best,

Julian
 
Roy,


I think with any adjustment its what the individual is comfortable with. There are lots of articles, websites and guru’s who drone on, it gets mightily confusing.

I think colour correction should be simple; Photoshop articles have complicated what was a very easy process.

All the stuff I do goes for high-end printing, I can spend hours (days sometimes) correcting images, which usually in my opinion the photographer should have taken into account when shooting.

The photographic images you see on billboards, high-end magazines are usually the result of 45 layer/channel plus retouch.

Personally, I’d stay clear of RGB, colour variation can be extreme.
I was taught the additive or subtractive method of correction, this does suit curves, when combined with channels, masks, it really does give you fine control.
Julian way is a good method that basically sets your tonal range.

The rules are simple 3-5% in the highlight, 85-92% in shadow. By creating a simple B/W test strip, with these percentages you apply to your images.

I always go by the numbers, rarely judge by eye. The golden rule is keep correction down to the minimum.

You have an advantage in the fact that your images are correctly exposed, all you need really to do is set the image cmyk to the correct enlargement/resolution and give it a sharpen.
 
Roy,

Depends on what you are trying to do with levels either color balance or increase the tonal range of an image.
I prefer curves to levels, since it gives me more control over both and it is the first part of my workflow.
If my image is a little too warm, I select the red channel in curves and adjust the curve to reduce the reds etc.
Generally, I use curves to set my black points and white points, which once done, takes care of balancing the image since the midtones fall into place as a result. Here’s my workflow, the steps of which may be common knowledge, but hopefully helpful. It’s more simple than it seems once you have it set up.

In curves, click on the black eyedropper in the dialog box and when the color picker come up, set the values in the RGB checkboxes to 10, 10, 10. Close and click Click on the white eyedropper and set the same values to 242,242,242. Don’t touch the grey (midtone) eyedropper.
The reason for doing this is - think of a stop-wedge going from black to grey to white, the values of black starting at 0 and ending with white that has a value of =255. 0 equals blacks with no detail and 255 is “paper” white eg. an area that has no details, referred to as clipped or “burnt-out”.
Setting the black values to 10 and whites to 242 (or245) basically retains details in the whites and as much detail in the shadows (I found anything from a value between 1-10 in the blacks is not discernible in print, so no point setting it below 10). Some will argue, but I set mine this way and prints in magazines etc all come out great.
Close curves and when it asks if you want to save the new targets as defaults, click “yes”. This basically telling PS to map any true blacks in the image to a value of 10 and whites to 242.

Now, let’s say I want to quickly color balance a picture of an Ivory Gull. Open the info palette Window>Info, so it’s visible, from the tool menu, click on the eyedropper tool and as you move it over what you consider the darkest part of the image, or a true black, keep an eye on the readings in the info palette. Once you find the darkest spot (values reading close to 0) click to place a sampler at that spot. Odds are that all the RGB reading in the info palette will be close but not equal.
At this point, open up curves, click on the black eyedropper tool and carefully click on the sampler you placed on your image. It will make those RGB values 10,10,10. Since each channel has the same value, it basically renders the image neutral, thus color balanced.


Now, use the eyedropper tool again and move it over the whitest areas, or brightest areas and look at the info palette (values of 242-250). Once found, click a sampler in place. Open up curves, click on the white eyedropper and click on the sampler you just placed. All the RGB values will change to reflect 242,242,242. Since each channel has the same value, it basically renders the image neutral, thus color balanced.

NOTE1 – this works best on nicely lit images –images taken at sunrise or sunset will not benefit since this workflow will eliminate the nice warm colors that are desirable. Also, it is important to make sure you click on the darkest/whites parts. Clicking on a dark brown for example, or a dull white will not produce pleasing results.

NOTE2 – to find the darkest or whites parts of an image, open up Levels and start to move the black (shadow) slider inwards - hold the ALT (PC) down as you do this and the screen goes white. Move the slider until you start to see the first black/colored pixels appear. These are the areas of the image that are the darkest, or first areas to be ‘clipped’. Do the same with the white slider and the screen goes black. Move the slider until you start to see the first colored pixels appear. These are the areas of the image that are the whitest, or first areas to be ‘clipped’. Remember this area in the image so that you can quickly home in on that area instead of hunting for the brightest, darkest area. Then follow my instructions.

best,

Julian
Thanks for that Julian. I must admit I have not used Curves that much so I will give it a go. Quite a lot of this I already do like using the Alt key in levels and setting the Black and white points to values other than 0 and 255 (I have been using 5 and 250).
I will have a play with your method in Curves.
 
I think some of the information quoted here is more relevant to print output rather than on-screen viewing. At work we have always produced 'flatter' looking images for litho reproduction. This is because the fine dots at each end (the white dots in the shadow areas and the black dots in the highlight areas) tend to fill in or get lost. By reducing the dynamic range of the image to 10-242 rather than 0-255 fine detail is preserved better at each end of the range. I don't think this is so critical for on-screen images.

The main problem with all these adjustments is that you are dependent on the accuracy of your monitor. An image can look great on your own screen but will look completely different on everyone else's. I also find that sometimes an image which I have processed in the morning for the Gallery looks completely different later in the day, when my screen has completely warmed up and the ambient lighting is different. It's a minefield!

Ron
 
I do think beauty is in the eye of the beholder. If you ask someone why they think a particular image is a good photograph 99 times out of a hundred they aren't going to rapsodise about the tonal range. They will be looking at the subject. Look at any list of top photographs and you will see lots of shadows, blown out highlights, centered subjects and other cardinal sins.

Photoshop has become a curse in many ways we all need to stop fussing about which setting to use for which adjustment and get out there and take photos!
 
Photoshop has become a curse in many ways we all need to stop fussing about which setting to use for which adjustment and get out there and take photos!
I am sure that only those that are interested in the subject will bother to even read this thread - you may not particularly like/enjoy the processing side of things or playing around in PS but I do which is why I started the thread.
Sorry if it has offended you but no one has forced you to read it. Live and let live is my motto, we all have different interests.
 
Sorry if it has offended you

No offence taken at all!

I can spend sizeable parts of my working day sat in front of Photoshop preparing photos and graphics for publication and knowing that most of the work won't be noticed by most of the folk that see them.
 
Mono says it well...at the end of the day, some of the hours we spend on an image for print won't be noticeable to most people.

I prep my images based on outputting them for print at 11 x14 size I optimize my images as best as I can to give me the best print at that size. A web-jpeg is easily saved from this - I can't think of having two processing workflows, one for web and one for print since I have to produce the best image I can with an output size in mind. Print is forgiving, especially when the halftoning procedure during the offset printing softens any slightly "on-purpose" oversharpened images.

Roy - I think 5 for blacks and 250 for whites is fine, depends on what printer can hold what minimum date in the shadows and highlights. I go by numbers in the Info palette, since my monitor hasn't been calibrated for a while.

All the stuff that appears in print (say Birdwatch) has come out looking nice and true to my monitor view.

JH
 
I think that Roy's right in expressing the importance of postproduction, However, Mono really does have a point about Photoshop.

We have a problem here in the UK with digital because of general overcast conditions we get, every time we venture out armed with a camera. Photoshop can get mind-blowingly tedious correcting these images.

It does come down to the individual and how far they wish to proceed with the technical side.

Ron, summed it up precisely, all of us are to some extent dependent on the quality of the equipment we use, the light conditions of the room, time of day, and even our eyesight.

It’s true that these figures are based on print production, but that ultimately should be the final destination for your image. It produces the best results.

Julian is exactly right in preparing two versions one for print the other web.
 
It does come down to the individual and how far they wish to proceed with the technical side.
That was exactly what I was trying to say Pe'rigin - I am one of the few who actually enjoys photo-shopping so for me there is no question about it being tedious.
I go out every day walking and taking snaps weather permitting but there are enough days in the UK when you just cannot get out so I also find plenty of time for playing around on the pc.
 
Roy,


I think you're the lucky one living where you do in this country, the light does seem to be cleaner and crisper.


On a technical side, people do get confused whether to adjust their image. I do get asked a lot to advise them. Again it's down to the individual how far they go.

The easiest way to judge, which does get me out of prolonged explanations is as follows.

Take an image which has a good (daylight) tonal range.

Convert to CMYK

Go to channels

Turn off CMY

Look at the black (K), if you can see a complete ghosted representation of the image then your image will need adjustment. The % will be the amount -3,or 5% in the highlight measurement.

What you want to see is the Black (K) covering about 3/4 of the image with areas of clear.

How to adjust and what areas need attention is entirely down to the individual and method of choice.
 
levels 'n stuff

pe'rigin, I wasn't following exactly why you would want to go that route to check the tonal range? I think what you're saying is the same as I did above. Looking at the numbers in the info palette, which gives a measurement of the tones in the highlights and shadows (regardless of whether the screen is warmed up or not), takes the guesswork out of adjusting tones/colors by eye. That's the beauty of the numbers - they don't lie, but monitors do :C
One other important point for me, is that since images are captured in RGB, it doesn't make sense for me to convert them to CMYK and then back again. Once you convert to CMYK, the numerical data that govern how the color is represented on your monitor are usually changed. Going backwards and forwards from RGB to CMYK to RGB isn't recommended.

The only time images should be converted to CMYK is when images are commercially printed AND you have been given a specific CMYK profile to use in your conversion (eg. Uncoated Sheetfed, Fogra 27 etc). A good printer will be able to advise you on which CMYK profile to use in your conversions.

If you are asked to "convert to CMYK" without any other information from someone, you are essentially taking a gamble. Without this specific profile from a prepress provider, you are converting your RGB file "blind" and hosing the integrity of the image. Photoshop converts to whatever CMYK profile you manually set in the Edit>Color Settings panes. RGB pane should be set to Adobe RGB (1998), CMYK (in the UK) to a specific FOGRA profile (dependant on paper stock either a coated or uncoated,etc). Everyone should ensure that Color Settings are set up this way to avoid issues down the line.
If you are having stuff printed in the US or ASia, then you will have to change the CMYK from FOGRA to another profile since printing specs for color are different than the FOGRA specs.

If at all possible, let the printer convert your RGB files to CMYK. If they will charge you for this (often best to ask them), then you can submit them as CMYK to save money, but only after they tell you what profile to use in your conversion.

I think I may have gone a little off topic ;) but this is important for those that may get involved in having to prep images for print.
 
Julian,


I don’t think that you’ve gone off topic; it’s a good point you’ve made.

Colour adjustment is a very confusing subject that is never explained properly by people who get paid to advise or write about this subject, some very badly. To many people out there do not understand technical processes.

The clicking of a shutter is only the start of a process that will go through various stages to a finished image.

But, as I have written to Roy, it really is down to the individual how far and when they wish to adjust.


We view imagery, light projected on a glass screen or plasma monitor in RGB, with the best calibration available the colours are only simulated. It’s an ideal platform for viewing colour only.

Other factors affect the situation as Ron pointed out, room light ambience, the position of the monitor against a light source, etc.

No process prints in RGB. We can’t reproduce these colours; only view them over the web.


Using the figures I’ve said in CMYK gives the perfect tonal result, for reproduction and viewing, which will fit the majority of printing processes. It is technically correct as it set to ‘blue light’, what all of us have above us, a blue-sky light source.

If people do send images away for publishing, then learning or having a basic knowledge on reproduction would be advisable.

I certainly would not trust any magazine in colour corrections, for the simple reason, apart from expertise, they don’t have time.
 
No process prints in RGB. We can’t reproduce these colours; only view them over the web.
Very true. There are plenty of RGB colours which are not reproducible in CMYK, royal blue, for example. This colour is a bit of a nightmare for us, as it is very popular for school uniform sweatshirts and keeps popping up in school magazines.

I think for many people on this forum, who post images in the Gallery, the differences between online display and printed output will only become apparent when they take one of their favourite screen images to be printed and are disappointed when the on-screen brilliance becomes lost in the process. As I understand it, on-screen RGB images are additive (where the red, green and blue combine to produce white), whereas printed CMYK images are subtractive and progressively filter out the reflected light from the paper to ultimately produce black. I suppose this is irrelevant to the original topic but I'm a bit bored at the moment.

Ron
 
Warning! This thread is more than 14 years ago old.
It's likely that no further discussion is required, in which case we recommend starting a new thread. If however you feel your response is required you can still do so.

Users who are viewing this thread

Back
Top