Lighting for Digital 5

Dave Montizambert searches for optimum digital exposure

We’ve heard it a million times and here it is one more time: Ansel Adams' golden rule for optimum capture in his B&W Zone System is, 'Expose for the shadows and develop for the highlights'. This makes sense for B&W film where all contrast control can be applied in the darkroom after the exposure. However, it makes no sense with digital capture; if you expose for the shadow you run the risk of pushing the highlight tones over the edge into white without detail. Now once all detail is lost in the highlight, and I mean really lost – 255 levels in all three channels – there is nothing to work with, no chance of rescue with Lightroom or Adobe Camera RAW’s highlight recovery slider and, since it is digital, no resurrecting silver halide crystals hiding deep in the emulsion. Furthermore, there is less chance of these fixit methods turning out with images processed 'in-camera' (ie JPEGs) than there is with RAW images processed after the fact in your favourite RAW processing software – in-camera processed JPEGs contain less data than RAW files and so highlight recovery is more unlikely.

With RAW digital files the number of stops from middle grey to pure white is less than the number of stops from middle grey to pure black. In fact where detail in the shadow ends is a matter of some debate, it really comes down to how much noise and banding of tonal gradation you are willing to risk. In addition to this, digital favours the highlight, it assigns more bits to describe the highlight than it does to describe the shadow. Don’t know what a bit is? See side bar entitled 'Bits, Bitmap, and Bits Per Channel'.

Dramatic backlit lighting: Look at Image [1] of model Sarah Madro. She is illuminated by light reflecting off a 1.3 x 2 metre white nylon fabric stretched over a panel frame placed in front of her on the camera left side of the image frame – see lighting diagram Image [2]. The origin of the light energy for this main light source comes from a strobe/flash mono-block place behind her on the camera left side of the image frame. This light origin also provides heavy dramatic back lighting for Sarah giving her separation from the dark background. I often use this one light dramatic backlit technique indoors on rainy days to simulate outdoor sunlight. To fill in the shadows a second reflector panel was placed to the camera's right side of Sarah, this fill light source caught stray light from the backlight and redirected it onto her dark side. This fill reflector was placed further away than usual to create darker shadows.

Most professional D-SLR cameras today create 12-bit encoded data. 12-bits describe or break down the greyscale into 4,096 discreet steps or levels of brightness. An increase of one digit in bit depth, say from 12- bit to 13-bit, will double the possible levels of brightness. Very recently Canon announced their new 1Ds Mk III that they claim to be 14-bit. If this new Canon is a true 14-bit camera, its captures should have in the neighbourhood of four times more levels of brightness to play with than does a 12-bit camera! See Bit Depth/Tone correlation in Image [7].

A greyscale has pure white at one end and pure black at the other end. In between these two extreme tones we have 4,094 levels of brightness with 12-bit, that is to say that the intermediate tones of the greyscale are broken down into 4,094 separate steps or levels of tone. We are using slight variations in brightness to simulate a continuous tone greyscale and if the steps of that progression are fine enough, we can fool the eye into believing that it is seeing a continuous tone progression.

In fact we don’t really even need this many levels to create the illusion, it can be done with considerably less. Your printer dumbs down bit depth to about 6-bits for printing and when you work on your image in Photoshop at 8-bits or when you create a JPEG image, you are only using 256 levels of brightness to describe the whole greyscale. 8-bit is considered low-bit and 10-bit and above is considered high-bit, at any rate more bits equal more levels of brightness which equals more editing head-room – an important factor when processing images as well as correcting, or manipulating, files in Photoshop.

As I mentioned earlier, digital assigns significantly more bits to the highlight than the shadow. Expressed in brightness levels, digital applies 2,048 levels to the first stop down from pure white at the highlight end of the greyscale (see top graphic in Image [3]). This is half of the available 4,096 levels possible from the 12-bit encoded data. The second stop down uses half again of the available levels, 1024. The next stop 512, the next 256, the next 128, leaving only 64 levels to describe the sixth stop in the shadow.

Sixty four levels sounds pretty bad, but it is more than enough to describe a shadow. However, at 64 levels you may run into trouble if you try to lighten up the shadows to show more detail. Many digital photographers play it safe, under-exposing their captures to ensure highlight detail, then in their RAW processing software or by opening their image in Photoshop, they use the appropriate controls to brighten the image to bring back the mid- and shadow-tones. This method may be a necessary evil in uncontrolled lighting situations, but be aware that if you go too far with ‘playing it safe’ you can wreak havoc with shadows. When you lighten a file in your image editing software, you are pushing the distribution of bits further up the greyscale, in other words if you lighten the file by one stop, where you once had 64 levels of brightness in the shadow at the sixth stop, you now have half (see bottom graphic in Image [3]). In addition to this, noise, which typically resides in the darker tones, will be emphasised. If you wish to maximise the available bits your camera has to offer, it makes more sense whenever possible, to place the brightest significant highlight as close to burnout as possible and then during processing place the shadow, quarter, mid, and three quarter tones to taste.

In Image [1] of model, Sarah Madro, the shadows are intended to create a dramatic look and so are quite dark. If you had a change of heart after the fact and decided to lighten the shadows of this 12-bit capture by one stop using curves or levels, the sixth stop shadow tones would only have 32 of the original 64 levels (see bottom graphic in Image [3]). In Image [4], we see a before and after close-up of where the transition from lighter shadow into darker shadow occurs in the cheek to hairline area. In the lightened version, we see this shadow gradation breaking apart (over-emphasised for teaching purposes) – the subtle gradations from one tone to the next are too far apart to create the illusion of continuous tone. This bit of ugliness is called posterising or banding and is as undesirable as cracker crumbs in bed. If you are not sure how dark you want your shadows, then I suggest a higher fill ratio during lighting and then darkening the shadows after the fact during RAW processing or in Photoshop with curves.

Page 1 - Page 2

Photo Quote: The camera is no more an instrument of preservation, the image is. - Berenice Abbott