Fix it now or fix it later?

For every photographic problem that might be addressed at the time of shooting, there always seems to be someone’s glib response that you “can fix it in post.” It is indeed possible to do many things with Lightroom, Photoshop, or GIMP – but is that the best or easiest way to do it? Let’s examine nine common correction operations, how they play out when shooting or in post, and which seems to be the better (or at least most efficient) option.

1. Perspective correction and leveling. Using a wideangle lens (<35mm) at anything but a dead-level position causes converging (or diverging) verticals. In the dark days before Photoshop, converging verticals were mitigated with PC lenses that shifted the lens relative to the film. This shifted the horizon and the effective viewpoint of the camera (10mm of shift compared to a 24mm frame height can move the horizon line more than 40% up or down). Older shift lenses had larger image circles to accommodate this, but they also show chromatic aberration on digital sensors – and they required inconvenient stopped-down operation for viewing and then metering. Newer lenses have electronically controlled apertures that help compensate for some of this. Correcting converging verticals in post-processing avoids the optical compromises and difficult metering, though the “warp” to the frame (which goes from rectangular to trapezoidal) cuts down the frame size, changes the effective aspect ratio of the picture, and compromises fine details if you’re starting with a low-res file. But the bigger problem is that most programs are not really capable of correcting perspective issues without distorting the vertical/horizontal proportions of the picture – generally making things look too tall. DxO Viewpoint has a ratio corrector, but it still requires visual estimation of a viewing angle that you never saw in real life). In terms of misery level, the easiest option is to get a wider lens, get as close as you can keeping the subject level, and simply crop as necessary. Time of shooting.

2. Vignetting control. Older lenses, especially symmetrical ones, often exhibit darker corners on digital sensors (they did on slide film as well, but on the negative film that most people used, this was less visible). Vignetting is a limitation imposed by physics. It also occurs with lenses designed for digital, but in many cases the camera can automatically compensate for a known lens when generating a JPG. At the time of shooting, when generating a RAW file, you basically have only a center filter as a choice. These very expensive filters impose big losses in terms of film speed (typically requiring 1.5x the exposure) and work best at smaller apertures. Even where there is no Lightroom profile for your lens, other solutions such as CornerFix and Adobe Flat Field allow you to shoot control pictures for repeatable corrections in the future – and to shoot with no exposure increase. Post.

3. Fill light. There are those who profess never to use flash and only whatever light is available. No one knows what they do with pictures that exhibit dark eye sockets, awkward shadows, and dominant light sources that point the wrong way. You can fix some of this in post, but simply the raising the exposure in certain parts of the image can make it difficult to maintain a natural-looking result. The major solutions here are to compose to face the dominant light source, use a reflector, or (heavens forbid) use fill flash. Time of shooting.

4. Light balancing (cooling). Low incandescent light provides unique challenges for digital sensors, almost all of which have noisy blue channels. Room light is typically pretty low, and the ISO setting on the camera typically ends up being pretty high, which means more noise across all channels. Using white balancing to compensate for reddish incandescent light exacerbates the problem in the blue channel by amplifying it even more. If you have a steady enough hand to do it, using a 80A (KB-15) filter drops the red and green channels so that the noisy blue channel is not unduly amplified. You lose 2/3 of the light doing this, but it cuts down on chroma noise. Time of shooting.

5. Light balancing (warming). The red channel does not suffer from the noise issues that the blue does. So it is ok to amplify it later. This in itself is not too compelling, but consider how at the time of shooting, warmer, no matter how warm, seems better – and yet in editing, things often look too warm. So consider limiting your filter filter use to an 81A (or KR3) and do any additional warming later. Post.

6. Red enhancement. The didymium red-enhancing filter has largely gone out of production (possibly due to demand and possible due to RoHS considerations). Its effect, which is to suppress “every other color” in the red-yellow range and then everything else past it, is extremely difficult to reproduce in post, if only because the peaks and valleys, occurring every 25nm or so, do not correspond with available adjustments to color in Lightroom (many of these actually fall between colors). Although it might ultimately be possible to reverse-engineer the effect, it would be a pain… Time of shooting.

7. Graduated neutral-density filtration. In color work, at the time of shooting, your only real choice to make the sky darker without a polarizer is a graduated neutral-density filter. The best versions are rectangular and allow you to rotate and move the horizon line. That said, they are much more unwieldy and flare-prone than circular grad filters, which are compact, easy to use, but completely inflexible in horizon line (midpoint of the gradient) placement. And with either, the hardness of the gradient needed is defined by the lens in use (oddly, only the rectangular versions offer a choice of hardness). Longer lenses require a harder cut. Provided that the dynamic range your scene permits it, the better solution is using gradient filters in Lightroom. These are variable for center position, rotational angle, and steepness of the gradient. In fact, they can be combined with other adjustments. The quality loss is minimal for simply darkening part of a scene; usually it is a relatively detail-free area like the sky. Post.

8. Specialty filtration. Softeners, diffusers, cross-screens, diffractors, and the like are filters for which there is no good Photoshop equivalent (assuming, of course, you are into the looks these filters create). Time of shooting.

9. Black and white tone adjustment. If you are into the effects of colored contrast filters on black-and-white film, you cannot very easily bolt such a filter onto a camera with a Bayer filter, because some filters (particularly red) can cause havoc with demosaic-ing. The Channel Mixer function in Photoshop (and Lightroom) lets you selectively raise or drop colors (at least within -20/+20) without too deleterious an effect on the image. The sole exception is the Leica M Monochrom, which having no color data to work with, must be filtered at the time of shooting. Post.

10. Correcting mixed lighting. Balanced fill flash falls apart any time that a flash is being balanced against something with a different color temperature. The most common problem is in room light, where at base ISO, flash essentially becomes the only light source, making the subject bright but the rest of the frame relatively dark. Raising the ISO tends to even out brightness, but it leads to pictures where the background is yellowish and the flash-lit subject looks normal. Although this can be corrected with a lot of work later, the easiest thing to do is to gel the flash with an 85A filter to make its light the same color as the room light. Time of shooting.

None of this is to say that there is anything wrong with post-processing digital images, and in fact, some things can only really be done digitally (fine-tuned and synchronized white balance, distortion removal, sharpening, etc.). But it is to say that a little more care in shooting can cut down on the time and frustration involved in post-processing.

# # # # #