HDR FOR ADULTS

High Dynamic Range imaging (HDR) was invented to archive light, but was unfortunately hijacked early on to create lurid imagery that has little to do with photography and a lot to do with bad taste and judgment. The pair of images below, processed from the same image file, illustrates what can go wrong…

Fortunately, we see less and less of the exaggerated effect as time goes by. Here’s what it’s really for — skilfully handling scenes that span a large range of light without getting over-excited. Professional HDR processing is all about solving the technical issue of a scene’s dynamic range that’s too great for even a modern cameras sensor to handle (and they improve all the time). Overwhelmingly, the main situation is an interior, especially one with a view toward the outside, like this…

[caption id="attachment_162554" align="aligncenter" width="1680"] Colonial black-and-white house, 25 Malcolm Road, Singapore[/caption]

Now, this looks perfectly reasonable and not at all high in contrast (high dynamic range and high contrast are to all intents and purposes interchangeable), but the reasons it looks this way is that it was carefully processed. In fact, it was shot in four frames each 2 ƒstops apart.

HDR, in fact, refers to the process of compiling different exposures of the same shot, NOT to the result, though unfortunately, sloppy terminology has meant that now for many people it has come to mean a style of processing. Let’s get this misunderstanding out of the way quickly. It was invented for CGI (computer-generated imagery) in the movies as a means of calculating how the light falls in a real scene so that it could be reproduced for such CGI elements as a flying Superman or whatever other character had to blend seamlessly into a live-action plate. It was then adapted for the use to which we photographers now put it.

The HDR part of the process involves simply shooting a series of identical frames with different exposures, sufficient to capture everything from the brightest highlight to the deepest shadow, typically 2 ƒ-stops apart. Software then combines them into a single image file. Here’s a worked-through example, using an interior that was extreme in dynamic range. This was in Thailand, with dark wood panelling shot entirely in natural and ambient light in the middle of a very bright tropical day in sunshine, and it took 6 frames to cover the range, spaced 2 stops apart. As you see, the darkest shot has no clipping, even in the extreme highlights, while the lightest shot renders the darkest shadows (under the table) as mid-tones. It’s important to cover both these ends like this…

[caption id="attachment_162464" align="aligncenter" width="1680"] Khampun House, Ubon Ratchathani, Thailand[/caption]

Vary the shutter speed, not the aperture, so that the content of the image doesn’t change. The settings here were ISO 100 and ƒ16, then 1/8sec, 1/3 sec, 1/1.3sec, 2sec, 5sec, 13sec. In fact, it was a little more complicated than that, because I’d chosen to use a shift lens, rotating it to cover a larger area for a large file size, so what you see here is a section of the finished image.

The next step is even simpler, combining the range of exposures into a single image file. There is plenty of software to do this, but what I use is Photoshop’s Merge to HDR Pro. Simply load the images, then choose to create a 32-bit floating point tiff. This is important, as we’ll see in a moment…

The critical part of the procedure is the actual processing, because while at this stage you have a single file that combines the light captured from several, and you’ve effectively archived all the light in the scene, making that look reasonable on an 8-bit monitor means making compromises. How do you want the image to look? Do you really want to show everything? At this point, we should talk about the ‘look’. Specifically about what I call a ‘photographic look’. There’s no established terminology about this, but the processed ‘look’ of a photograph is a new phenomenon, since the invention of new digital ways of handling brightness and contrast on a local scale. In film-based photography it had no meaning, because all film and photographic paper responded in essentially the same way to exposure and processing. Photographs had a consistent look. Digital algorithmic processing of Raw files has changed this. In Adobe Camera Raw, which is shared by Photoshop and Lightroom, three sliders in particular work on the entire image but in a local way: Highlights, Shadows and Clarity. They use TMOs (Tone Mapping Operators) that alter brightness by means of very localised contrast. Essentially, they search around their pixel neighbourhood and make adjustments based on neighbouring pixels. The results are powerful in that they can ‘open up’ shadow areas (any areas, in fact) with increased localised contrast and crispness, but applied strongly produce an effect that is hyper-detailed — more like an Andrew Wyeth painting, for example. You can do this to an ordinary image, but it’s more tempting to do to an HDR image simply because it contains more dynamic range.

However, what I need in my work — as do most professionals — is an image that actually looks like a photograph, and that means turning away from TMOs. There are many ways of processing HDR, but making a 32-bit floating point TIFF file and then processing it with local adjustments is solid and reliable. The example here uses Photoshop, which offers the 32-bit option and continuing in ACR (Adobe Camera Raw, the normal Raw processing engine) with a single click. The experience is notably different from using ACR on a single Raw file, because it’s as if you had captured the scene with a super-sensor. The Highlights and Shadows sliders can be pushed quite far without the typical tone-mapped look described and shown above, while local adjustment, such as with radial filters, works incredibly powerfully. The secret is to hold all this processing power in check by moderating your own taste. Here’s the sequence for part of the finished image, which you’ll see at the end…

[caption id="attachment_162491" align="aligncenter" width="1680"] Step 1: as opened, before any adjustment[/caption]

[caption id="attachment_162500" align="aligncenter" width="1680"] Step 2: Use Highlights and Shadows sliders, which have a less extreme effect on an HDR file than they would on an ordinary Raw file[/caption]

[caption id="attachment_162509" align="aligncenter" width="1680"] Step 3: Raising the Contrast counters the inevitable mid-tone flatness of the Highlights and Shadows sliders. Small adjustments to the Whites and Blacks (which set the white and black points)[/caption]

[caption id="attachment_162518" align="aligncenter" width="1680"] Step 4: Lower Saturation, which always tends to rise with an HDR file[/caption]

[caption id="attachment_162527" align="aligncenter" width="1680"] Step 5: Use radial local adjustment filter to fine-tune[/caption]

[caption id="attachment_162536" align="aligncenter" width="1680"] Step 6: Continue to make local adjustments[/caption]

Here, then, is the final image, which actually covers a large range…

[caption id="attachment_162473" align="aligncenter" width="1680"] Khampun House, Ubon Ratchathani, Thailand[/caption]

HDR processing in this way underlines the wider issue of the ‘look’ of an image in modern digital photography. Anything is possible, but because of that, there’s arisen a largely unthinking response among many photographers to open up all shadows indiscriminately and reveal as much as possible. Processing tools can be applied to the entire image (that is, global) or to selected areas (local), The attraction of global processing is that it seems on the face of it neater and simpler, and is certainly quicker. The arguments for local processing, however, are first that this is a precision approach, and second that you can stick to traditional tools that will better preserve a natural photographic look. In any case, local processing is almost always performed on top of a relatively simple base global setting.

Simply put, if you have the time and are concerned about refining the processing to bring out the very best in the image, first make a base setting for the entire image that looks good overall with the exception of some specific areas. Then use local adjustment (such as a radial filter) for one, two or many specific areas, tailoring the settings just for that. This way you can target with great precision areas in which you may want to lighten, darken, change the contrast, colour or saturation, and anything else. Moreover, you can temper the use of those algorithmic TMOs for your base global settings and stay in control of the Look. For anyone with a passing interest in old-fashioned wet darkroom printing, this was how skilled prints used to be made, though the clock was running and extreme precision wasn’t quite possible. The tools to use depend on personal preference, but a radial filter has a lot going for it, because if you get the degree of feathering right, the adjustment effect ramps smoothly and imperceptibly.

Here’s another, last example of an interior that covers a much larger range than it appears to. But then, surely that’s the mark of good processing — you don’t notice it.

[caption id="attachment_162545" align="aligncenter" width="1680"] Casa Roman, Calle Real, Manga, Cartagena, Bolívar, Colombia.[/caption]