Exposure blending is an editing technique to better control global contrast in high dynamic range scenes where either the shadows or highlights might be clipped. Specific parts from multiple exposures are combined to insure that there is no (or significantly less) clipping in the final image. Exposure blending recovers shadow and highlight detail. While there are a number of algorithms and apps available to create HDR blends, the consensus seems to be that the results don’t always look natural. Shadows might be excessively light, global contrast looks to be off, and there may be halos around high-contrast edges.
Manually blending exposures using luminosity masks generally overcomes these problems. Luminosity masks focus on specific tones in the image and then seamlessly blend or taper into other tones. This is an ideal characteristic for exposure blending. However, using them for this purpose is usually not as easy as simply applying a luminosity mask to one of the exposures. Additional modifications to the masks, the exposures, or the manner in which the mask is applied to the image are often needed.
I don’t do a lot of exposure blending with my images, but from what I’ve done and have seen others do, there appears to be three main goals:
- Recover clipped highlights and shadows. This is essentially a working definition of exposure blending. A high dynamic range scene where the sensor is unable to capture either the brightest or darkest elements needs to use at least two exposures blended together to display the full dynamic range of the scene.
- Eliminate or reduce noise in the shadows. Sometimes the sensor can capture a scene’s full dynamic range, but the shadows are quite noisy. Blending in a separate shadows exposure that has the shadow values shifted to the right on the histogram provides both more detail in the shadows and less noise.
- Create realistic contrast in the blended image. This would include local contrast in each exposure that is blended as well as global contrast in the blended image.
One thing to NOT do with exposure blending is to try and finish the image based solely on the blending process. The goal should be to make a blended exposure that is a good starting point for additional development in Photoshop. In fact, it probably makes sense to aim for a less-finished blended image as there are lot of useful development techniques that might not be available during the blending process.
Method #1: Match exposures
For many landscape images, exposure-matching is a convenient blending method. It makes creating a perfect blending mask (the luminosity mask) nearly foolproof. This is the method demonstrated by Dave Kelly in the video below. The important step in this method is making the light and dark exposures to be blended look pretty much the same in Lightroom or Camera Raw. This often involves increasing the Exposure of the dark image and decreasing the Exposure of the light image. Additional adjustments to Highlights, Shadows, Blacks, Whites, and Contrast can then be incorporated to make the dark and light images appear similar. Clipped values will not be recoverable, of course, but much of the rest of the image can be adjusted to have similar brightness and contrast. The matching is especially important in the transition zone, where the dark and light exposures “meet.” A good exposure match in the transition zone means that there will be no blending halos in areas of strong contrast (like along the horizon or the edges of buildings).
When using matched exposures for blending, it’s very easy to create a mask that works to facilitate a perfect blend since the images already look quite similar. However, it’s still important to modify the mask to make sure it brings through the best-exposed pixels from both the light and dark exposure. For example, dark values should come from the light exposure in order to decrease shadow noise in the blended image. Mask modification usually involves creating some pure black and pure white areas via a Curves or Levels adjustment of the mask or painting black or white directly on the mask. It’s just the transition zone that needs to have various shades of gray in the mask in order to blend together pixels from both exposures.
NOTE: If using smart objects of the RAW files for blending, after the blending mask has been created, consider revisiting the RAW files by double-clicking the smart object thumbnails and creating better tone, color, and/or contrast in the areas revealed by that exposure. Just remember to try and maintain the exposure match in the transition zone.
Method #2: Paint through luminosity selections
While exposure-matching tends to work well for landscape images with separate sky and foreground areas that have an obvious transition zone, more complex subjects or more complex lighting situations might require a different approach. In the video below, Emil von Maltitz uses various luminosity masks generated with the TK8 plugin to create selections that then serve as stencils for painted masks. The painted masks reveal various image elements from different exposure layers. In Emil’s example, four different exposures are used, and very little work is done on the RAW files before exporting them to Photoshop for blending. It’s the mask-painting through luminosity selections that creates the blend.
The TK8 plugin makes it easy to find the right mask, but the real key for making mask-painting work, I think, is to use a lower-opacity brush (Emil uses 20 to 40 percent opacity) and then judiciously choosing where to paint on the image to bring through the desired exposure. Multiple brush strokes with lower-opacity brushes allow the effect to be slowly built up as each brushstroke adds additional paint to the layer mask. Using more than two exposures for blending adds to the complexity of the blending process, and Emil uses groups for helping to keep the Layers panel organized. Painting through luminosity selections also requires keeping track of what parts of each exposure layer will be useful in the final blend. Emil obviously has a good sense for this, though it would likely take some practice to be proficient when using more than two exposures.
SUMMARY: Exposure-matching and mask-painting are two methods for manually exposure blending high dynamic range scenes in Photoshop. Both methods employ luminosity masks to achieve a seamless blend. Exposure-matching has the advantage of creating an excellent transition zone so that critical areas, like horizon lines, have no halos in the blended image. Mask-painting allows for using original exposure information and is ideal for bringing out details and textures in the blended image.
PERSONAL NOTE: When working with high dynamic range scenes, I almost always start by applying a linear profile to the RAW image and then clicking the “Auto” button in Lightroom or Camera Raw. This process is the best way to see what’s actually recoverable in any given exposure. The linear profile does a fantastic job of recovering highlight details. And, since it has lower contrast in the dark areas than the Adobe profiles, it often shows better detail in the shadows as well. High-contrast shadows with Adobe profiles can look nearly black. Using a linear profile plus “Auto” helps to see what’s really there.