This is a technique I use on a lot of my images to try and show the full dynamic range of a scene as is visible to the human eye. Digital camera sensors are very good at what they do but they are no match for our eyes when it comes to the range of brightness values they can detect and record. Various figures are quoted to compare camera sensors with eyes in terms of dynamic range (an internet search will keep you entertained for hours), but the bottom line is that where we can see details in deep shadow and in a bright sky within one scene, the camera will record black shadows and a white burnt out sky. This is one of the main factors that leads to people being disappointed with their photographs, and rightly stating that the photo is nothing like what they saw at the time.
There are various techniques available to help deal with this problem, one of the most popular in recent years being HDR software programs like Photomatix and the Merge to HDR tool in Photoshop. In experienced hands, HDR software can produce excellent results, but it has also led to a proliferation of some very strange (even ugly) looking landscape images on the internet. That's fine if your aim isn't to create images that are very true to the original scene, but for a lot of landscape photographers, realism is a primary goal. American photographer Joseph Rossbach describes HDR images as 'grungy' looking – it's the same sort of look you get if you overuse the shadows/highlights adjustment in Photoshop. I should repeat that in experienced hands the results can be great, but for now I believe I can get better results using the more hands-on approach described here. So I don't use HDR programs in my normal workflow, but that's not to say I won't in the future or that you shouldn’t.
My approach to manual blending has been borrowed from a lot of other photographers and involves bracketing the shots I take on location, then selecting the parts of each bracketed shot that best represent what I saw and combining these in Photoshop using layers and masks. There are some excellent video tutorials on this process on the internet – for example, check out Peter Eastway's tips at www.betterphotography.com. There are also some great tips to be picked up on this and other photographic techniques on the Australian Photography Forum at www.photoforum.com.au/forum.php.
In the field
All good digital SLR cameras and a lot of compact cameras have an exposure bracketing function. This allows you to automatically take a series of shots of the same scene at different exposure values with a single press of the shutter button. A simple example that is easily set on most cameras is to take three shots, one at the correct exposure as indicated by your camera, one underexposed by 1 stop, and one overexposed by 1 stop. Different cameras allow more than three bracketed shots to be taken, a wider exposure range of the bracketed shots, and different exposure increments to be set, but that's the basic idea.
Three bracketed shots are usually more than enough for manual blending. I vary how much I under and overexpose the bracketed shots depending on the scene. For low contrast scenes, as you might see very early or late in the day or on very dull days, increments of 1 stop are often sufficient – if a scene is very contrasty, I space the exposures apart as widely as possible (eg. 2-3 stops), and even that is sometimes not enough to capture the full range of brightness levels. I regularly check the histograms on my camera to see if the highlights or shadows are being clipped, and if so, alter the exposure range and shoot again. If there's a problem, it's most often at the bright end of the spectrum due to a bright sky or reflections – when this happens, I take an extra shot at a much darker setting and again check the histogram for any clipping.
Graduated neutral density filters can do a lot to even out exposure differences between a bright sky and dark foreground, but they have their limitations. I use them quite a lot, but they can't easily deal with an uneven horizon line that is broken by trees or hills, and will do nothing to solve the problem of bright reflections on water (although a polarising filter will help here). Long exposure shots of waterfalls and cascades will also often create problems as areas of white water quickly burn out (ie. disappear off the right hand end of the histogram) before the rest of the shot is adequately exposed. I'd encourage photographers to use GND filters because they can reduce the time you spend at the computer later, but they won't solve all your exposure problems.
In the RAW Converter
I assess the bracketed RAW files in SilkyPix Pro – for most people it's more likely to be Adobe Camera Raw or Lightroom, but the process is the same. By looking at the histogram and using the clipping warning, check that you have a full range of brightness levels represented – sometimes you will need all three bracketed shots, often two is plenty. To keep it simple, I'll presume we only need two shots to encompass all the highlight and shadow detail we want in the final image – one for the sky and one for the foreground. Set the exposure levels of the two shots to show the sky and foreground at their best, make any other adjustments to the RAW files you want, then take them into Photoshop. I save them in SilkyPix Pro as 16 bit TIFF files and then open them in Photoshop.
An alternative approach to bracketing your shots is to take a single shot at an average exposure setting, and then process the RAW file two or three times at different exposure levels. This can work well and is more economical in terms of saving space on your memory card and computer hard drive, but it is more limiting than bracketing shots. While you can push the exposure level of RAW files up and down quite a bit without creating problems, you can only push them so far. If you take a single shot that includes an area of deep shade, and then in the RAW converter increase the exposure level by one or two stops to bring out some detail in that shade, chances are you will also bring out quite a bit of unwanted noise. That can be fixed up to a point with noise reduction software but you're creating extra work for yourself, and applying noise reduction generally leads to some degradation of image quality. There is more latitude when it comes to reducing the exposure level of RAW files without creating problems, but the best quality image will come from a RAW file that is processed at or close to the exposure setting at which it was captured. I try and limit pushing or pulling the exposure level of a RAW file to less than 1 stop, although I will sometimes reduce it by up to 2 stops if I really have to.
This is where the fun starts. By the way, most of what I say here specifically applies to Photoshop, but you can do much the same thing in Photoshop Elements by using the layer mask from a levels or saturation adjustment layer and grouping the mask with your image layer using CTRL-G. There are lots of internet tutorials that describe how to do this – for example http://graphicssoft.about.com/od/pselements/qt/layermasks.htm.
Now that both images are open in Photoshop (Figure 1), copy the darker one (exposed for the sky) and paste it on to the lighter one (exposed for the foreground) – you'll see them sitting one above the other in the layers palette. You can stack them the other way around if you like, but I always put the darkest image on top for consistency. I’ve assumed you’ve taken the bracketed shots using a sturdy tripod so they will be perfectly aligned – if not, both Photoshop and Elements have alignment tools to get the layers to line up with each other.
Figure 1 - the two images opened in Photoshop
There are a lot of options for how you go about the next step. The final aim is a composite image with a well exposed sky and foreground, but this can be achieved using the eraser tool, the brush tool, channel selections, other selection tools, the Blend If tool in the Layer Styles dialogue, and probably others I'm not familiar with. The eraser tool is generally not a good option because it permanently removes pixels from the file – if you later decide to alter the way the images are blended together, you'll have to go back to the original RAW files and start again. A better approach is to use a combination of selection tools and masks - that way no pixels are removed and you can go back and make changes whenever you like. I've only just discovered Blend If and it looks like it could be useful in some cases (see http://www.adobepress.com/articles/article.asp?p=686520), but for now I'll describe how I use selections and masks.
With the darker image visible on your screen, you need to select the sky so you can create a mask that lets it show through onto the lighter image. If you have a nice clean, straight horizon, like in a seascape or desertscape, the quick selection tool will usually do a good job. Click and run the mouse pointer around the sky and everything will be automatically selected – if you've missed some bits, click on them to add – if your selection spills below the horizon, hold the ALT key and click on these bits to remove. Alternatively with simple horizons, you can just create a blank layer mask and paint into it with a soft brush to hide the bits of the image you want to remove rather than worrying about selections. The colour selection wand and the magnetic lasso will deal with slightly more troublesome horizons – you'll need to experiment with these and play with the settings to fine tune your selection. All selection methods will produce some artefacts around the edge of the selection, so play around with the 'feather edge' and 'conctract/expand selection' settings to see if they help. Some careful brushwork on the mask, using a soft brush at low opacity, will also help tidy up messy edges.
A method I've found to work very well in most cases, but especially when the sky is broken up by foliage like in my example, is to use channels to make your selection (not available to Elements users I'm afraid). With the top (dark) image selected, click on the channels tab and you'll see individual boxes for each of the red, green and blue channels, as well as the combined RGB channel. If you look at the individual channels, there will usually be one that shows the greatest contrast between a light sky and a dark foregound – most often it will be the blue channel. Grab the channel with the most contrast and drag it down into the dotted circle at the left of the bottom toolbar (Load channel as selection). Marching ants will appear around the selection, but don't worry if you appear to have missed bits of sky or included too much foreground at this stage. Click back on the combined RGB channel, then click the layers tab.
Whichever method you’ve used to select the sky, the next step is to go to the bottom of the layers palette and click on the little circle inside a square (Add a mask) to create a mask. Elements users will have to refer to the tutorial I mentioned earlier. If you've done everything right, you'll see the foreground from the lighter image show through below the properly exposed sky. You usually need to tidy up the mask with a soft brush set at low opacity, so with the mask selected, paint black over the foreground, being extra careful as you get up towards the horizon, and paint white over the sky and possibly bits of foliage to darken areas that look too light.
No selection methods are foolproof, and you need to practice with them all so you know which is best to use in a given situation. But now that we have a good blend of sky and foreground, most of the work is done. There are likely to be parts of the foreground or midground that look a little overexposed now – like the white water in my example. These are easily darkened by selecting the mask you've created and painting over the problem areas in white with a soft brush – set the brush opacity low (eg. 15-20%) so you can build up to the brightness level you want in small steps, rather than overdoing it with one swipe of the brush at high opacity.
When the overall exposure of the composite image looks balanced, add a levels layer and check the shadows and highlights clipping warnings (hold down ALT while you click on the black and white triangular pointers below the histogram in turn). This will tell you if further adjustments to the mask are needed – if so, it's just a matter of painting black or white into the mask to lighten or darken the problem areas.
The blended image in my example (Figure 2) is much closer to how the scene looked in real life than either of the RAW files. Some final adjustments using levels and curves and the job is done (Figure 3). The whole process, including processing the RAW files, took me about 10 minutes.
This basic method will allow you to be as pedantic as you want about the exposure levels in all parts of your your image. If you use enough layers and spend enough time, you can basically recreate what the HDR programs do automatically – but then you might as well be using an HDR program! But with practice, basic manual blending is a simple and surprisingly quick method that will give you a lot of control over your images and produce very true-to-life results.
Figure 2 - the blended image
Figure 3 - the final result