Lack of latitude with digital capture

Table of contents

1 Dealing with digital capture's lack of latitude

2 Digital versus film

3 What you'll need if you want to try this

4 In the camera

5 On the computer

6 About the Author

[edit]

Dealing with digital capture's lack of latitude

(or, when the shadows are OK, the highlights are white!)

[edit]

Digital versus film

Most amateur photographers are convinced of the advantages of digital photography for very good reasons. We can experiment and take lots of shots without wasting anything, except perhaps our time. We can take control of the image through software in a way unimaginable in film, and infinitely more quickly and more easily. The costs of materials are lower, and now even high end hardware is not very different in price from film equivalents.

Most of the problems of early digital have been overcome, particularly in the area of resolution, and any sensor of 6MP or above is good enough for just about anything that 35mm film was good for. But there is one area where film still scores heavily against digital and that is in latitude. Film had what is described as a 'shoulder' in its response curve. As more light hits the film it doesn't respond linearly, and so it is able to 'hold' the highlights and still provide detail in parts of the image that are over-exposed. Digital can't do this, and any over-exposure very quickly 'whites out' leaving ugly and unconvincing plain white highlights, and bleaching the colour, especially from skies and clouds.

I'm going to describe a technique that I use to help get round this. As by and large a landscape photographer, I just got fed up with beautiful skies that came out anaemic when they should have been vibrant. And when I got the skies right, everything else was hopelessly dark and gloomy.

[edit]

What you'll need if you want to try this

A camera that can shoot in RAW format
A RAW converter (I'm using Pixmantec's RawShooter Essentials which is free)
An image manipulation package such as Photoshop (I'm using the GIMP which is also free)
If your camera can bracket (take pictures at different exposures) automatically, so much the better, but it's not essential
A reasonably big memory card to hold the multiple RAW images (say 1GB)

[edit]

In the camera

Set your camera to shoot in RAW. The reason for this is that later we're likely to need to 'push' the exposure in the RAW converter so that the shadows aren't too dark. This will cause some degradation of the image, but RAW is much less affected by this than a JPEG would be.
Next, if your camera allows it, set up a bracket set with one exposure at the metered level, and one about 1.5 stops less.
If your camera has a matrix meter or similar overall method of exposure control, use it. Spot metering is not useful for this technique.
On my Nikon D70 I also set the camera to continuous shooting mode, which means that every time I press the shutter, the camera automatically takes two exposures, one at -1.7 stops and one at the metered value. But of course you can simply take two shots, changing the exposure yourself between times, if your camera doesn't do things automatically.

[edit]

On the computer

Download your RAW files in the usual way.
Open your RAW converter software, and use it to open both of your exposures of the image.
RAW can do a lot to compensate for under-exposed shadows, but very little to help with over-exposed highlights. Depending on the scene you have captured, you might choose the under-exposed image, or the 'correctly' exposed image. But if the metered image has blown out the highlights so that there's lots of an undifferentiated pure white or drained colour, you'll have to work on the under-exposed image. Nine times out of ten, if your image has a bright sky, you'll be working on the underexposed image. In the example here, I've used the under-exposed shot.
We're now going to use the RAW converter to process the image twice. Look at the highlights. Because this image was deliberately under-exposed, there might still be a bit of headroom, and you might be able to adjust the exposure upwards a bit to brighten the highlights. After all, they are supposed to be highlights, not just a bit brighter than the rest! In this case I didn't need to. So you can see in the screen shot on the right (this is RawShooter's correction tool box) that the Exposure compensation slider is set to 0.00, and that the blue channel highlight is close to the maximum, but not clipped. On the other hand you can see that the shadows are far to the left of the histogram (too dark) and that the lowest tones are clipped.
Convert the image the first time, and save the output file.
Process the file again, this time ignoring the highlights, and concentrating on the shadows and mid-tones. In this case, as you can see from the screenshot on the right, I have 'increased' the exposure by 2.5 stops using the Exposure compensation slider. You can also see from the histogram at the bottom of the screen shot that green and blue channels are both seriously clipped, and indeed the blue channel is virtually invisible at the right hand end. By contrast, we now have much better mid-tones and shadows with very little clipping at the lower end. Take a look at the two thumbnails in the RawShooter screenshots. In the first one the image is far too dark overall, and the street details are murky to say the least. But in this image we have a beautiful blue sky and clearly defined clouds. What about the second thumbnail? Here the picture looks generally clearer, and the street is well enough lit. This is how most cameras would meter the scene. Most of the image is well exposed. But look at the sky! It's white and uninteresting, and the clouds have disappeared completely. It's also not a faithful description of the scene. I was there, and I know. But if you just saw this image, you wouldn't have any idea of what you were missing. You might think it looked fine enough, but so much would have been lost that you weren't even aware of.
Convert the image a second time, and save another output file. Make sure that your converter doesn't simply copy this one over the first one! You may need to 'save as…' to make sure you end up with two files.
Now we're going to get the best of both worlds by using an image manipulation program to combine the two images in one.

Open both of your images in your image manipulation package. I'm using the free GIMP software but I'm sure you'll be able to work out what to do in Photoshop or other package you might have.
Work first on the underexposed image, and use the free select tool to select the well exposed highlights. This doesn't have to be especially accurate, but make sure your selection includes the edges of the underexposed parts of the image. See the screenshot below.

Next you need to 'feather' the selection so that there will be no harsh boundary in the final image. How much feathering depends on the image, and only experience will tell. Start with 400 pixels. You will end up with a nice smooth boundary, which was why it wasn't important to be very precise when you first did the selection. It will look like the screenshot below.

Now you will copy this selection, and paste it into the other image. But before you do this, make a duplicate layer in the second image so that when you paste into it, you will have the original unaltered image in the bottom layer. This is a good insurance against things going wrong!
When you paste the selection it will become a 'floating selection' and it's most probably going to be in the wrong place. Don't panic! It will look like the screen shot below.

Next you will need to move the selection until it fits. This can be quite tricky, and you must do it at 100% enlargement or it will be impossible to get it accurately aligned. Not quite right….

… Ah, perfect!

Now you need to anchor the selection to stop it moving. In GIMP this is done by clicking on the little anchor sign.

You should now have an image with two layers, one which is the original with the overexposed parts, and a second layer which has the transplanted highlights from the under-exposed image. If you flick between the layers by turning on and off the visibility of the duplicate layer, you'll easily be able to see the results. Below I've shown the duplicate layer and its histogram, and then the original layer with its histogram. You can see how the second histogram shows the blown out highlights, whilst the first shows the blue channel peak well below the clipping level.

I hope you'll agree that the first image is much better than the second. You may well recognise this image from my gallery on this site, and this is exactly the technique I used.

For this image, it might seem that it's a lot of trouble to go to for a bit of blue sky and a couple of clouds. Well, only you can say whether you think your images are worth it!

[edit]

About the Author

Steve Brown lives and works in London, and has moved from film to digital over the last two years. His main photographic interests are architecture and the French countryside. More of his images can be seen in his TLF gallery (http://www.thelensflare.com/profile.php/sgbrown) and on www.hautevienne.moonfruit.com.

Retrieved from "http://www.thelensflare.com/docs/index.php/Lack_of_latitude_with_digital_capture"