UNDERSTANDING DIGITAL EXPOSURE:

Part 1: Exposure Reduction for Highlight Retention
Part 2: Digital Exposure & Noise
Part 3: When Blocked-up Shadows Aren’t Really
Part 4: Take a Balanced Approach to White Balance
Part 5: Dynamic Range
Part 6: Extending the Tonal Range
Part 7: What’s the Real Difference Between RAW and JPEG?

GENERAL:

1) COST-EFFECTIVE PHOTOGRAPHY
2) CONTROL THE RANGE OF FOCUS
3) IMAGE INTERPOLATION
4) LOSE THE DEAD SHOTS
5) SCANNING 35MM FILM
6) THE RAW vs. JPEG DEBATE…
7) IS PHOTOGRAPHY EASY?
8) MUST EVERYTHING HAVE AN ADOBE SLANT?
9) A CAMERA TO PAINT WITH
10) WHAT'S THIS COMPOSITION THING ABOUT?

Understanding Digital Exposure

1: EXPOSURE REDUCTION for HIGHLIGHT RETENTION

A recent advertising campaign by Fuji went like this: “Ever feel like your digital SLR is missing something? If you’re fed up losing highlight detail because of the limited dynamic range1 of most digital SLRs then maybe it’s time you took a look at...”

Maybe it is. But what can we do about this problem, other than buying Fuji products?

Digital exposure is something of a balancing act because sensors are so sensitive to the highlights. If we can’t balance the range of tones by other means, we will have to reduce the exposure time so less bright light reaches the sensor. But this may also mean some shadow detail will be lost and midtones will be darkened slightly.

If there’s one golden rule that applies to typical digital exposures it’s this: be sure not to waste space to the right of the histogram. Some highlights should be totally white, like brilliant reflections, but others can be pulled back by exposure reduction. However, there is always the risk that the important highlights will be pulled back a little too far showing an empty space to the right of the histogram and we may also see information at the left that runs off the display altogether. This probably means shadow detail has been lost. So when appropriate it’s best to fill that right-hand gap with data.

If you have time to look at the histogram after you take a shot, do it. Then adjust if necessary. Checking the histogram is very important, but it's not always possible or convenient. A lot depends on the situation. You won't hear of too many tripodless photojournalists missing a shot because they were studying their histograms for gaps and clipping. Reduced exposure suits this style of shooting.

Apparently the DSLR's LCD histogram display is based on the camera's JPEG settings. This is worth remembering if you're shooting RAW and regularly checking the histogram for exposure accuracy and, of course, highlight retention. If we really want to get picky let's not forget that nicely filling up space to the right of the histogram may in fact blow data in a colour channel. Underexposure may do the same at the other end. In fact it's a good idea to take a close look at a histogram in software to see how it compares to the camera's version. LCD histograms are typically too small and therefore less reliable.

If you’re using a compact camera it may be best to sacrifice some shadow detail. Smaller sensors typically have less space on which to fit light receptors (photosites or pixels), so they’re made smaller. This narrows the gap between the brightest and darkest tones the sensor can record. Larger sensors have more space for bigger receptors and can record a wider range of tones, but the difference isn't huge.

When you haven't brought your tripod, to get a shot at all some decisions have to be made about exposure and its relation to depth-of-field. In this case, even with an image stabilising lens (G9), it was necessary to zoom in and take an exposure reading from the pale, setting moon. Then the lens was zoomed out again until camera-shake was no longer a problem at the fastest aperture and lowest possible ISO setting (80). The main feature of this image is the moon, so the exposure is based on it, but having the branches a little bit sharper would have been an improvement. (Image unedited.)

Smaller sensors also mean more noise – the digital equivalent of grain (see next page). So it’s more difficult to control shadow detail in image-editing software which, in turn, is why it may be best to settle for compact camera exposures with slightly darker shadow detail than we would like – it may even be lost. If you try to enhance these darker details you may aggravate the noise.

However it’s important to realise that noise looks worse on-screen than it will in a print, so it’s best to experiment to see how far you can go with your particular camera. Viewing distance affects the visibility of noise. On-screen assessment at 100% magnification should take this into account. This is an important point. Get a print done first. You may be pleasantly surprised. Don’t look at your quality widescreen TV up close. You won't be impressed.

Higher ISO ratings will also increase noise levels especially if there's a gap needing fixed on the histogram at the bright end (typically caused by underexposure). In fact, with compact cameras the results can be very poor indeed. This means it’s best to shoot at the lowest ISO possible. Anything beyond an advanced compact's ISO 100 setting is usually best avoided, especially if editing is required and overall image quality is important.

When it comes to digital exposure we find that experts and self-styled online gurus appear to be divided. Imagine that! Some recommend marginally increased exposure with RAW and slight underexposure when shooting JPEGs. The thinking is you'll get more highlight data from RAW and less chance of noise in the shadows; each JPEG is camera-processed RAW data and because information is "thrown away" it won't do so well in the highlights.

There are photographers who will recommend outdoor RAW underexposure and indoor RAW overexposure. OK... And some (like me) will tell you to err on the side of caution by dialling in slight underexposure in most handheld situations because there will be less risk of blown highlights.

Then you'll read that it's easier to get detail from the shadows in slide film than it is from a digital image. But the next day you'll go online and learn that digital images have impressive latitude compared to slide film, but in the shadows only. Hmmm...

Handheld digital exposures often need help to control highlight data. The original image, top-left, was slightly underexposed to favour the highlights. Even so, the histogram tells us that highlights and shadows have clipped – some detail has ben lost at both extremes. The version above-right shows the image after selective shadow enhancement. But lifting the shadows amplified the noise. In some areas this was unacceptable so noise was reduced in software. The difference is subtle but effective (see left where darker strips are the original).

Many amateurs avoid reduced exposure because of the risk of noise, but the final image shows that the results can be very good indeed. Without compensation the camera would most likely have lost too much highlight detail. This JPG image was shot with an advanced compact at ISO 80. In a situation like this a low ISO image from a DSLR would show no signs of shadow noise.

Above: deliberately reduced exposure with an advanced compact has held subtle tonal information in the petals. The camera's exposure is from a light reading based on mid-tones – in this shot the sky is a mid-tone. But leaving exposure calculations to the camera can result in finer details being lost. These small areas of significant blown data will most likely not show up in the histogram. The challenge here is similar to shooting small bright flowers against a dark green background when it's likely the camera will overexpose blowing important highlight detail. Exposure should suit the main subject and make allowances for essential bright tones. Mid-tone and shadow enhancement is usually possible in software, especially with DSLR images shot at the lowest ISO setting. In this context a RAW image would retain slightly more highlight information than a JPEG but both could be processed for similar results.

One thing's for sure: digital cameras lose highlight detail at the drop of a hat. But I'll doff my hat to someone called Tom Judd who said: "With slide film, we try to avoid overexposure and the resulting 'blown' highlights. Shadow detail may be lost, but that is usually less objectionable. The goal with the digital camera is the same – avoid blown highlights. But, as with negative film, the final image is created in the (digital) darkroom." That makes a lot of sense. We have to do the best we can. We hope we won't be forced to compromise by clipping data either way. But it's a challenge, shot to shot. And software manipulation is part of the solution.

In an ideal world we should use cameras with large sensors and the most powerul processors, always shoot at the lowest ISO possible2 and expose with the significant highlights in mind. Whatever cameras we use, we must accommodate the highlights. But some are wary of this approach because it can mean expanding shadow detail too far in image-editing software, increasing and/or amplifying noise. There are also issues with weakened colour strength caused by reduced exposure.

But surely the main problem is the extent of the exposure reduction. How much is too much? Like it or not, digital exposures need to be based on lighter tones. So if we are to rely more on exposure reductions to hold the highlights we need know by how much we can expand the significant shadows in software without visibly harming the image.

The results with DSLRs in particular, especially full-frame, are very good indeed. The example below is another deliberately extreme example. Please remember that. The only purpose here is to show how flexible digital shadow detail can be. (This image is barely useable.) In situations where the exposure has been intelligently reduced to hold on to significant highlights, we can expect very useful results that don’t damage data by amplifying noise. And don’t forget that a print will compress the range of tones, so careful shadow expansion can be helpful.

It’s very important to remember too that shadow detail in a normally exposed shot should usually be targeted in software. Some make the mistake of applying corrective adjustments to every darker pixel in the frame. This will probably impair other areas of the image rather than enhance them. Choose the areas that need most attention. Single them out for manipulation.

The top image is the original JPEG straight from the camera, the bottom is after shadow enhancement. Below we see the curve that was applied and the resulting difference in the histograms:

The image below shows 100% detail:

The final image (below) shows the results of editing the RAW version of this image. It simply wasn’t possible to control all the tones using Canon’s RAW-editing program. But if two images were processed, one for the highlights and one for the shadows, the combined result would be much better than the curve-processed JPEG. Bad mistakes can sometimes be corrected very successfully if shot in the RAW format, but that's no excuse for bad technique.

1 The extent of tones captured by the camera's sensor and utilised by the processing unit.

2 Some cameras offer ISO 50, but there's no such thing. The DSLR design threshold for exposure is actually ISO 100. ISO 50 is really ISO 100 with a bit of in-camera manipulation that some claim marginally helps to suppress noise. But according to the manufacturer there’s about 1 stop less dynamic range in the highlights at ISO 50. Typically, the last thing we want to do is aggravate a sensor's inherent highlight limitations. But this contrived ISO speed comes into its own when shooting wide apertures in bright conditions or with flash. It's also a bit like a ND filter that restricts light allowing for longer exposures and shallower depth of field.

home | Personal Pages