One area I often see where misconceptions arise is the exposure of raw images in-camera. Even some professionals believe they should “protect the highlights” as they used to do when exposing film, and underexpose the highlights instead of “exposing to the right” in your histogram as I did in this “froggy” shot. Half the raw file tonal levels your camera can record are in the brightest f-stop! Your camera is simply counting photons in a linear fashion at this stage, with no application of a gamma factor (unlike film). If the camera is just counting photons, it makes sense that the brightest f-stop will have the most pixels assigned in the system – remembering that an f-stop is an opening to light – and since f-stops following the brightest stop have to diminish by a factor of two in brightness (photons) this affects the number of pixels assigned. So as we progress towards the shadow end, each “stop” of data is being halved, leaving far fewer brightness levels assigned to darker tones. In a 12 bit raw file (4096 total pixels), the brightest stop would have 2048 pixels assigned, with the following f-stop 1024 pixels, the next 512, the next 256 and so on down to the shadows with ever decreasing numbers of pixels by a factor of two. Therefore shadow details are more prone to noise and lack of detail as we progress downward. So capture the maximum highlight tones to start. A larger highlight capture means more pixels for the shadows in a raw file (bearing in mind that the sensor has a limited number of total pixels). However, the caveat to this advice is that any “blown” or “clipped” highlights (rgb 255, 255, 255) will contain no information and translate in print to the colour of your paper. So be careful to watch your histogram for spikes on the highlight end. Blown pixels are gone for good. To elaborate, once the highlight data is clipped in a RAW file, no curve, exposure adjustment, or highlight recovery can bring back real tonal gradations — they simply aren’t there. Pulling a curve down under 255 just gives you a flat white area that becomes flat gray.
A related area of confusion for some concerns “Picture Styles” as they are called in general. These are settings you make in-camera and can be somewhat misunderstood. Because photographers in general know that raw files are “developed” in software settings/sliders for output to some other final file format (TIFF, PSD, JPEG) it is understood that the raw file remains intact and untouched by these software settings or as Canon calls them in DPP 4 “recipes.”
However, because sliders are set to zero when raw files come into processing software, one could be forgiven for thinking nothing has been done to the raw files as yet. However, we are looking at a jpeg gamma corrected representation of the raw file at this stage, and even before in camera, as Bruce Fraser explained. The crucial point being that the “Picture Style” that was set in camera is a manufacturer specific set of patented instructions or “recipes” if you will, which are ALREADY baked into the jpeg representation of the file. These include the colour science I referred to in my previous blog – gamma, noise reduction, lens correction etc. So the various “Picture Styles” (named in Canon cameras Landscape, Portrait, Neutral etc.) have already set up a definite look before you even start post-processing. This can be helpful or a hindrance depending on what your final processing goal is. For instance in Canon cameras “Landscape” will have more contrast and sharpening and strongly saturated greens and blues, but “Portrait” softens contrast / sharpening and ramps up reds for more pleasing skin tones.
A related point re: jpeg representation of the raw file being baked in, is that Bruce Fraser told us that this usually involves a strong “S curve” being applied to the raw data. A consequence of this means that your jpeg may show the highlights as being blown when they actually may have 1/2 to a full stop of “headroom” – the highlights in the raw file are not really blown. So depending on the brightness range of the scene, there may be more room for an exposure boost to capture every possible brightness level in the original “linear” raw file capture. If the brightness range of the original scene is below twelve or so f-stops (the average brightness range capture today for most modern cameras), there is likely present some extra potential for more exposure. The point being that capturing the maximum amount of brightness levels will also produce a file with less noise and more detail (tonal levels) in the shadow areas.
****Update and Credit****
Re: my reference to Bruce Fraser – my first edition of this post had a link to a paper by the well known late raw file guru Bruce Fraser in which he explained the raw file – and some would say for the first time – to a broader audience. However, the link disappeared. It was originally on Adobe’s website. If you can find it by all means have a look, as I did. The core info in this blog is straight from Bruce Fraser.