Listening to the Light – Examples in Image Development
One of the hardest things about developing an image in Photoshop is figuring out what changes to apply to improve it. This is an acquired skill, I think, that, like composing an image with the camera, improves with practice. Sometimes I know exactly what I need to do in Photoshop, but I also spend a lot of time going back and forth on different possibilities, experimenting with alternate techniques, and generally exploring options before settling on a particular adjustment. Even then, I may go back and readjust an adjustment layer depending on what happens with subsequent layers.
Image development is a dynamic process, not a one-way street. I receive questions asking which technique to use in developing an image and when to use it, and the answer is always the same: It depends on what the image needs. Responding to the image often means that a standard “workflow” probably isn’t going to work. It’s really the image dictating what needs to be done, not the photographer or some repeatable steps. It’s necessary to respond to the light in the image in the same way you respond to the light in the field. Don’t try to control it. Instead, work with it, listen to it, and let it determine the direction of development.
There’s a corollary to this concept of listening to the light that is very important. Once you do figure out what the image needs, you have to have the right tool or technique to address the issue. As such, I thought it might be instructive to occasionally write a post on how I think about what an image needs at different points in its development and discuss my approach to solving the “problems” that I perceive. In other words, what I heard when I listened to the light, and how I responded to what it was saying. The goal being not to go over every layer in the development of a particular image, but rather to look at a few of them and attempt to discuss the collaboration that was occurring between the image and me at that time. I’ll say up front that it’s not always easy to describe exactly what I was thinking. Sometimes it’s just intuition and sometimes just a fortunate experiment that helps to move image development forward. However, I also think that instances like this maybe reflect our deeper understanding of the light and that we just don’t always have the vocabulary to communicate these feelings.
The image below is one that is partially developed. It shows a sandstone detail that I recently photographed in twilight shortly after sunset. I know abstract images can be confusing, so just to get past the perceptional difficulties here, the camera is pointing downward to capture the layers of texture and color as they recede from the camera. Four Photoshop layers were added after this point in development. I’ll go over them one by one explaining what was going on and my reasons for adding the layers.
The first layer added was a luminosity painting layer, which involves dodging and burning through luminosity mask selections to lighten or darken specific parts and specific tones in the image. It’s described in detail in this tutorial. My goal in luminosity painting is generally to create an evenness of light so that the colors and textures assume a large role when viewing the image. This often involves lessening some of the natural shadows and highlights in the scene that might draw the eye unnaturally in the print. Dodging (lightening) with luminosity painting can sometimes remove excess saturation in dark areas and can add texture in light areas. Burning (darkening) can improve color richness to areas that look washed out and can remove distracting highlights that draw the viewer’s eye.
I usually start the process by painting through a Basic Mid-tones selection since it can be used to both lighten and darken the image, but I also use other selections or combinations of selections from masks of both the Darks-series and Lights-series depending on what I want to accomplish. Once the even light starts to come out in the image, the uneven places seem to stand out a bit more. In this way, luminosity painting sort of becomes self-directing. It’s just a matter of looking at the image and seeing and addressing the unbalanced light and then painting through an appropriate luminosity selection to correct it. I usually have all the Darks- and Lights-series of mask sitting on my Channels panel, as well as a Basic Mid-tones mask, so I can quickly grab what works best. Below is the luminosity painted Burn/Dodge layer for this image. The mouse rollover shows how the image looked before luminosity painting to provide a better idea of where paint was applied to even out the light. (Note: For all the the rollover images, it might take a short time to load the second image, but continue to hover the mouse over the image until it appears. Once it does, you should be able to move the mouse quickly back and forth over the edge of the image to see the change.)
Blacker areas on the luminosity painting layer darken the image, whiter areas lighten it, and 50% gray areas are unchanged. The luminosity selection being painted through determines which pixels receive paint and how much. The luminosity painting layer is generally one of the most dynamic in the layer stack. I continually return to it as subsequent layers shift the light, and apply additional paint, black or white, to bring back the evenness. Below is how the image looked after luminosity painting. The rollover shows the image before the luminosity painting layer was added so you can readily see the difference between before and after.
Once I had the light somewhat balanced with luminosity painting, the next thing I noticed was the lack of saturation in the colors and an overall flatness to the image. While luminosity painting can be used effectively to increase local contrast, using the technique to even out the tones across the entire image can result in a loss of general contrast. The benefit of balancing out the light almost always outweighs the risk of increased flatness though, since general contrast can be easily improved. Because the image was both lacking in general contrast and saturation, the most logical choice for addressing both issues is an S-curve on a Curves adjustment layer. One concern I had with this, however, was the fact that my histogram was looking pretty good at this point. An unmasked S-curve could cause loss of shadow detail as the dark tones got too dark, and the lighter tones might also start looking too light. I certainly didn’t want to clip any shadows in this image taken in soft light, and I generally try to avoid strong whites. S-curves also can really punch up the saturation, sometimes out of proportion to the desired contrast change. A mid-tones luminosity mask (Basic Mid-tones is always a good one to try) would be useful in this case for confining the adjustment to just the mid-tones in order to keep the dark and light sides of the histogram unchanged and for providing a more gentle increase in saturation. The Basic Mid-tones masks for this image is shown below.
This mask is always a bit hard to “read” in that it’s overall grayness is confusing. Essentially what a Basic Mid-tones mask reveals is the image’s mid-tones. Both the light and dark colors are darker (concealed by the mask from being adjusted) and mid-tone values are lighter (revealing the adjustment). The lack of pure black and pure white in the mask makes it seem like it would be ineffective at concealing or revealing anything in a meaningful way, but it actually does a very good job of targeting adjustments to just the image’s mid-tones, as designed, sparing the dark and light colors from being overly adjusted. The histogram below demonstrates this. The S-curve adjustment that was applied on the masked adjustment layer is shown in the figure below.
Like luminosity painting, an S-curve adjustment through a Basic Mid-tones mask seems to find its way into all my images since it’s a really good way to adjust overall contrast without messing up the histogram. If the saturation looks right before the adjustment, I change the layer’s blending mode to Luminosity to avoid adding saturation to the image, but generally the boost in saturation, especially with a small adjustment like this one, looks right and improves the image.
Below is the histogram before the curves adjustment was applied. The rollover is the histogram after the Curves adjustment through the Basic Mid-tones mask. Notice how the two ends of the histogram remain fixed. The improved overall contrast comes entirely from expanding the tonal range in the mid-tones.
At this point I thought the on-screen version of the image was looking OK. I really didn’t see anything else that bothered me all that much. Whenever I reach this point of satisfaction, though, I make a print, and without fail, the problems I couldn’t see on my monitor are now quite obvious. I think this is because the transmitted light of the monitor makes almost everything look better. Pictures generally look better on-screen than on paper. So in this regard, viewing a print in reflected light encourages criticism, and it’s a useful tool, I believe, in helping to understand where the light in the print wants to go. The print in this case indicated a couple of problem areas, and the last two layers for this image are a direct result of viewing a hard-copy version of the image.
The most obvious problem was the blue color cast in some of the white rocks. This is completely natural given the light conditions at exposure—a clear sky with post-sunset glow in the west. However, in the print it seemed like the blue contrasted too much with the warmer elements in the scene. The image had become decidedly warmer as it developed, and these cool whites now felt a bit out of place. A warming filter would take care of this, but it wasn’t needed for the entire image and not even in all the white areas. Just the blue-colored whites need to be warmed. This is a situation where a painted mask works well. I created an Photo Filter adjustment layer with a warming filter and adjusted the filter until it removed the blue in the blue-tinted white areas. The adjustment is shown below.
The layer’s mask was then inverted to black to completely conceal the adjustment. A Lights luminosity selection was created to target the whites in the scene, and the marching ants were hidden. White paint was applied onto the layer mask to the blue-tinted white areas of the image. The lights selection helped target the paint to the lighter/whiter areas being painted. While this step was originally intended to remove the cool color in some of the whites, revealing some warmth in specific darker areas seemed to work well too. Because the Lights selection reaches into some of the darker tones, painting darker parts of the image turned out to be an effective way to warm-up other cool-toned areas as well. Below is the mask for the Photo Filter layer after it was painted showing the areas that were revealed to receive the warming effect. The rollover is the image before the warming was added to provide a better idea of how the blue-tinted areas in the image ware targeted for warming by the painted mask.
Below is the image as it appeared after the addition of the Photo Filter layer with the painted mask. The rollover again shows how the image looked before the Photo Filter adjustment layer was added. The blue-tinted areas in the rollover are now obvious and these areas look more coherent with the rest of the image once they’re warmed up, but again, it took viewing the print to see how this adjustment could benefit the image.
The final layer for this image involved color-cloning. While not all that bothersome in the on-screen version of the image, the unmatched rock color in the upper parts of the image, especially the upper right, looked unbalanced in the print. Revealing the Photo Filter warming didn’t help correct this A curve adjustment could be used to shift the color and then this color change could be revealed in the off-color rocks, but since there is plenty of the “right” color readily available in other parts of the image, an easier method is to simply paint in the desired color using the cloning color, retaining texture technique. Using this technique, the final layer is a pixel-containing layer set to Color blending mode. The desired color was sampled from other areas of the image and painted into this new layer where color change was desired. This technique worked particularly well for this image since the underlying color was slightly darker than the sampled color being painted into the layer. Color-cloning tends to lighten the painted area just a bit, so painting over a darker color keeps the area from appearing too light after painting. Generally when using this technique, I just paint the color onto the Color Clone layer at a low opacity to build up the color to the desired level. In parts of this image, however, I wanted to limit the color change to just the darker tones, so I painted through a Dark Darks and Shadow Darks selections to avoid adding the tint to the lighter tones. The image below is what the Color Clone layer looks like. Remember, this layer is set to Color blending mode so only the hue and saturation of things in this layer are transmitted to the image. The luminosity, which defines texture, comes from the composite of the layers below. The rollover shows the image before color cloning in order to see which areas were targeted to receive paint.
In summary then, four techniques were used to provide the finishing touches to this image: luminosity painting, a Curves adjustment through a Basic Mid-tones mask, a painted mask with a Photo Filter adjustment layer, and color-cloning. The different layers for these steps are shown below. Each technique was applied after assessing the image (listening) to see where change was needed, and each was specifically chosen to address the particular concern that was uncovered. As often happens, an actual print of the image revealed more than the on-screen version, but adding layers to correct the print also led to subsequent and noticeable improvement in how the image looked on the monitor.