Making the most of image channels
Tom Arah investigates Photoshop's best-kept secret - channels.
Undoubtedly the most powerful photo-editing control feature available today is the use of layers. When working intensively on an image, whether in Photoshop or Paint Shop Pro, you are likely to be handling multiple image layers along with more advanced options such as adjustment layers, layer effects and layer masks. The use of layers is so omnipresent you begin to wonder how users got by before their invention.
The answer is through the use of "channels". In fact it's a sure indication of a Photoshop veteran that they'll still be using the Channels palette almost as much as the Layers palette. More importantly it's a sure sign of a power user because channels still offer plenty of functionality that layers alone can't provide.
So just what is a "channel"? Unfortunately the technical answer - a 2D array of usually 8-bit information - is both less than illuminating and seriously off-putting. It's an idea we'll come back to but, before you stop reading, I think it's better to explore the two main types of channel - colour information and alpha - and to see how you can put them to use throughout your photo-editing work.
Colour Information Channels
The easiest way to see channels in action is through Photoshop's Channels palette. With a typical 24-bit RGB photographic image open what you'll see are four layers each with their own thumbnail marked RGB, Red, Green and Blue. Clicking on the top RGB layer shows the usual composite, full-colour image while clicking on any of the Red, Green and Blue layers shows a grayscale version of the corresponding channel. You can also access each colour information channel in turn with the Ctrl+1, 2 and 3 keyboard shortcuts with Ctrl + ~ returning to the normal composite view.
The full rainbow of RGB colours are built up through separate grayscale Red, Green and Blue channels.
Exactly what the Channel palette reveals is made clearer if you inspect the channels on a test file of a rainbow gradient. To begin with it's clearly significant that each channel is represented by a grayscale image in which each pixel can represent one of 256 values. It's also noticeable looking at the red areas of the rainbow on the Red channel that these are represented by the apparently empty white end of the grayscale. Looking at the yellow stripe of the rainbow across the channels it also becomes clear that this is present, ie white, in both the red and green channels while missing, ie black, in the blue channel.
Essentially what the rainbow image shows is how the full rainbow of 16 million possible RGB colours can be produced by combining just three Red, Green and Blue values between 0 and 255 (256 x 256 x 256). In our image, for example, the yellow stripe is represented by high values in the red and green channel and 0 in the blue channel. Rather than handling images pixel by pixel, Photoshop handles the colour information as channels, with the three 8-bit planes of Red, Green and Blue information combining to produce the final full-colour image.
Of course not all images are RGB-based, but that's not a problem as channels are very adaptable. Taking copies of our original photo we can use the Image>Mode commands to move the image into entirely different colour spaces. For Bitmap and Grayscale colour spaces there is only one colour channel limited to black or white and the 256 shades between black and white respectively. The Lab mode meanwhile is based on three channels, A and B which represent the image's chroma components between green and red and between blue and yellow and the L channel which represents Luminance. This separation of the image's brightness from its colour can be very useful - select the Luminance channel and then convert to Grayscale and the results are much better than if you convert to grayscale directly from a RGB scan.
Lab mode is occasionally useful but much the most important colour model after RGB is the print-oriented CMYK. Convert the rainbow test image to CMYK mode and the first thing you'll notice is a drastic colour shift as many pure RGB colours are outside the printable CMYK gamut. The second is that the image now has four independent colour channels - Cyan, Magenta, Yellow and Black. Look for the rainbow's yellow stripe and now you'll find that it is represented on its own Yellow channel and that here solid yellow is represented by solid black.
In CMYK images the channels represent the ink-based printing plates.
In many ways the system is very different to the RGB model as here Photoshop is having to deal with an ink-based subtractive model where all colours combine to produce black compared to the additive phosphor-based RGB model where all colours combine to produce white. In another way though the system, like all image modes, is very similar allowing Photoshop to produce any colour model's full colour gamut by juggling a maximum of four 8-bit channels each representing a maximum of 256 values (OK 65536 values if you choose the Image>Mode command's 16-Bits/Channel option).
Clearly channels are fundamental to colour handling and that's why it's a good idea to check the Channels palette to see how Photoshop is building each image's overall colour. This is particularly useful when dealing with the more intuitive combination of CMYK channels destined for print as here the RGB monitor-based composite simulation is not necessarily an accurate guide of what will be printed unlike the grayscale percentages on the individual channels. This is especially important with CMYK as press factors, such as dot gain and gray component replacement, mean that precisely managing the image channels and their resulting colour separation plates is crucial.
Channels aren't just useful for CMYK feedback, they can also be put to work - and in the more common RGB context. In particular rather than working on the composite image you can just as easily work on an individual channel simply by selecting it first. When colour correcting it's good practice to inspect each channel as this can often reveal a flaw in the scan, say a fuzziness or bias in the red channel, that can then be addressed directly using the Sharpness filter or Levels dialog. Alternatively, to produce special one-off effects, there's nothing to stop you applying an artistic filter such as the mosaic effect to just one channel. In all cases clicking on the View box of the composite channel lets you see the effect on the final image not just the current channel (remember to reselect the composite channel afterwards!)
More power is made available through Photoshop's ability to merge channel information. For producing special effects the Apply Image command lets you merge any channel information from another image of the same pixel dimensions into the current image's composite channel and lets you set a blend mode and opacity to control how the result is built up. Similar control is available in the Calculations command which also lets you specify a second channel and then merges the result either to a new channel, selection or new grayscale document. This can be used with the special Subtract blend mode to highlight which areas of an edited image have changed compared to the original.
More channel merging control was added in Photoshop 5 with the introduction of the Channel Mixer command and adjustment layer. This is for use on single images only and lets you control each channel by mixing in information from other channels. This can be used for special effects, swapping channels, and colour correcting scans. It's also ideal for creating high quality sepia or tinted images and, with the Monochrome option, for producing high quality grayscale images where you can control the contribution from each RGB channel to the final output.
Channel mixing is ideal when converting colour images to grayscale.
The colour control offered by the colour information channels is undoubtedly useful, but the most common use for channels is actually for managing selections. The separate colour information channels come in very handy here as often the elements of an image are far more clearly defined on one channel than on the composite. Isolating the bear in Photoshop's sample bear.psd image, for example, is far easier if you work on the blue channel.
That's great for making your selection in the first place but, once you've created it, you don't want to lose it. This is where the second type of channel come into play enabling users to save selections as so-called "alpha channels". This is done with either the Selection> Save Selection command or the Channels palette's Save Selection as Channel icon. Immediately a new channel appears with selected pixels saved as white (on), unselected pixels as black (off) and any anti-aliased or feathered pixels around the selection edge as intermediate grayscales.
Once the alpha channel has been created you can select it just as you do a colour channel, while double-clicking on it enables you to give it a more meaningful name and to set the overlay colour that will mark the channel if you turn on the Composite channel's view box. Crucially you can work on an alpha channel just as you can on the composite and colour channels. In particular this means that you can apply Level-based colour corrections and filters such as blurs. Even more powerful is the ability to locally edit the alpha channel using the brush tools so that you can paint in holes and inaccuracies in your selection.
Once the alpha channel is exactly how you want it, you can put it to use by turning it back into a selection using either the Selection>Load Selection command, the Channels palette's Load Channel as Selection icon or simply by Ctrl-clicking on it. If you have more than one alpha channel you can also create more advanced selections. If you have alpha channels for a figure and its surrounding background bush, for example, you can select just the bush by first selecting it and then using the Load Channel command to load the figure with the Subtract option. Keyboard shortcuts make this even quicker with Shift+Ctrl-clicking adding to the current selection, Alt+Ctrl-clicking subtracting and Alt+Shift+Ctrl-clicking selecting the overlapping intersection.
Alpha channels can be saved and loaded as selections.
Once you've turned your alpha channel into a selection it immediately appears onscreen as the famous marching ants and you're ready to transform it, copy it to a new layer or otherwise edit it. You should be slightly wary though as the marching ants aren't necessarily a completely accurate representation. In particular they only indicate those pixels on the alpha channel that are below a 50% gray threshold. For most solid selections this makes little practical difference but one of the huge strengths of alpha channels is that they aren't restricted to crude on-or-off selections but have 256 levels to play with which enables the creation of advanced variable opacity masks.
It's worth thinking about a couple of examples of how this variable masking capability can be put to use (both are actually a lot easier than they look written down). Create a new blank alpha channel with the Create New Channel icon, drag a gradient over it, reselect the composite channel, Ctrl-click on the alpha channel to turn it into a selection and now apply any of the artistic filters. The effect of the filter across the image depends on the grayscale values of the gradient mask so the end result creates the impression of your photo gradually turning into a work of an art.
Alpha channels can be used as variable opacity masks.
Or how about duplicating an alpha channel of a selection of text and then applying a Gaussian blur. By then subtracting the original alpha (Alt+Ctrl+Click) you can create a new channel that represents just the blur surrounding the text. Select this on the composite channel - or any of the individual RGB channels - and you can then use the Levels command to change the selection's gamma to produce a subtle or not-so-subtle glow effect - you can even delete the original text so that the glow seems an organic part of the photo.
Working in a similar way, and bringing in some lateral thinking, you can use alpha channels to produce a huge range of embossing, shadowing and vignette-based creative effects. When you are working intensively on an image in this way it's clearly important to be able to save your alpha channels with your file. Support is surprisingly thin however, which essentially means saving to either TIFF or Photoshop's own PSD file formats (and even here there is a limit of 24 channels per file including colour channels).
That's fine when working within Photoshop but alpha channels can also play a role in wider workflows when exporting images to other standard formats. A good example of this is when working with Web images. When saving images with the Save for Web dialog, any alpha channels you've created can be used as a mask to control JPEG quality or GIF "lossiness" and therefore file size. Create an alpha channel for the centre of attention in an image, for example, and you can keep quality high here while lowering it dramatically in the surrounding shadow areas. File size savings can be enormous at little cost in terms of end user perception.
Alpha channels also come in very handy for Web images when dealing with transparency. First create an alpha channel to mark off the area of your image that you don't want to appear on your Web version and then Ctrl-click on it to select it. Then call up the Export Transparent Image wizard and select the GIF option and you will find that the selected area has automatically been made transparent in the exported version (this saves making irreversible changes to the original). The wizard also offers the option of exporting to PNG which offers both full-colour 24-bit colour and 8-bit transparency. If you create a radial gradient alpha channel and then select that, for example, you'll find that your export file has a true vignette effect that blends into whatever background you place it over. Sadly support for 32-bit PNG files with transparency is currently thin on the ground with Director 8.5 being the most notable exception.
Alpha channels also play an important role when it comes to high-end print thanks to their ability to add spot colours to CMYK print. The easiest way to do this is to select the area of your image you want to have printed as a spot colour and then use the Channel palette menu's New Spot colour command. Using the Custom option in the dialog's Colour picker you can then select any colour from a range of spot colour libraries such as the Pantone set.
Adding spot colours in this way isn't quite as simple as it appears as you need to be very suspicious of the onscreen appearance as, by their nature, many inks are outside the RGB gamut and in any case Photoshop is simply simulating what it thinks the image will look like after overprinting (this can be controlled to some extent with the solidity setting). However, if you now save your file to DCS 2.0 format and load the composite EPS into a professional DTP application you'll be able to output separations for the four process colours along with any spot colour channel files you've created.
Spot colour channels enable ink colours outside the CMYK gamut to be handled.
All in all it's clear that channels have an important role to play in professional photo-editing from colour handling and selection through to Web imaging and high-end print. Having said this though it's important to recognise that time has moved on and for many purposes channels have now been superceded by layers. Once you've built up a selection, for example, it often makes more sense to copy it to its own layer so that you can work with it independently while keeping the original pixels underneath to fall-back to if necessary. The same is true of many other effects that in the past could best be achieved with channels but can now be handled non-destructively through adjustment layers, layer effects and layer masks.
In many areas then channels have been improved on - but certainly not in all. More importantly, they might have been superceded but they haven't been replaced. In fact layers couldn't exist without channels for the simple reason that layers are built on channels! If you think about it each layer is actually a collection of independent colour information channels with an alpha channel to mark transparency and an optional alpha channel to act as a layer mask. The same is true of adjustment layers which are effectively alpha channels through which a colour correction is applied.
Effectively then Photoshop - and indeed every photo editor - builds up the image we see by taking the resulting values from each layer's colour and transparency channels and combining them with the underlying channels' values layer by layer from the background up. The process is complicated by factors such as blend mode, opacity and adjustment calculations but effectively it's a straightforward step-by-step mathematical process. The images we see and handle mean nothing to the program itself, what it sees and handles are each channel's array of values.
Ultimately then the photo-editing application is always thinking in terms of channels - it's a habit that the power user would be wise to copy.
Hopefully you've found the information you were looking for. For further information please click here.
For free trials and special offers please click the following recommended links:
For further information on the following design applications and subjects please click on the links below:
[3D], [3ds max], [Adobe], [Acrobat], [Cinema 4D], [Corel], [CorelDRAW], [Creative Suite], [Digital Image], [Dreamweaver], [Director], [Fireworks], [Flash], [FreeHand], [FrameMaker], [FrontPage], [GoLive], [Graphic Design], [HTML/CSS], [Illustrator], [InDesign], [Macromedia], [Macromedia Studio], [Microsoft], [NetObjects Fusion], [PageMaker], [Paint Shop Pro], [Painter], [Photo Editing], [PhotoImpact], [Photoshop], [Photoshop Elements], [Publisher], [QuarkXPress], [Web Design]
To continue your search on the designer-info.com site and beyond please use the Google and Amazon search boxes below:
|designer-info.com: independent, informed, intelligent, incisive, in-depth...|
All the work on the site (over 250 reviews, over 100 articles and tutorials) has been written by me, Tom Arah It's also me who maintains the site, answers your emails etc. The site is very popular and from your feedback I know it's a useful resource - but it takes a lot to keep it up.
You can help keep the site running, independent and free by Bookmarking the site (if you don't you might never find it again), telling others about it and by coming back (new content is added every month). Even better you can make a donation eg $5 the typical cost of just one issue of a print magazine or buy anything via Amazon.com or Amazon.co.uk (now or next time you feel like shopping) using these links or the designer-info.com shop - it's a great way of quickly finding the best buys, it costs you nothing and I gain a small but much-appreciated commission.
Thanks very much, Tom Arah
[DTP/Publishing] [Vector Drawing] [Bitmap/Photo] [Web] [3D]
[Articles/Tutorials] [Reviews/Archive] [Shop] [Home/What's New]