Exposure Blending Images Containing Complex Horizons
Hi, I often run across a situation where I would like to use exposure blending using layer masks in photoshop to get a final properly exposed image from one underexposed and one overexposed image (e.g., one image sky is blown out and another where it is properly exposed but the rest of the image is underexposed). However, more often than not, the horizon in my photos contains objects such as trees etc. that result in banding/an unnatural halo in the area between the sky and the horizon. I assume that I am not the only person who has run into this issue. What techniques do you all use to overcome this obstacle? My issues may be that I am not using the most efficient selection techniques? I can post an example of my problem if you all think it would be helpful.
Visit my Nikonians gallery.
Visit my Nikonians gallery.
#1. "RE: Exposure Blending Images Containing Complex Horizons" | In response to Reply # 0
esantos Nikonian since 10th Nov 2002Sun 28-Apr-13 06:59 PM
That is a common problem when blending and that is why HDR processing has gained such a following. You can practice using different blending techniques and one can get quite good at it but sometimes the only real solution is to use HDR software and shooting multiple exposures. When I am out shooting and I know this might be an issue and I am not using ND Grad filters because of a very irregular horizon I'll shoot a series of exposures to capture the entire exposure range of the scene. Then when I am back at home I'll try either blending or HDR, and sometimes both, to see which gives me the best end result.
esartprints.com Ernesto Santos Photography
#2. "RE: Exposure Blending Images Containing Complex Horizons" | In response to Reply # 0
Antero52 Nikonian since 07th Jul 2009Tue 30-Apr-13 06:52 AM | edited Tue 30-Apr-13 06:55 AM by Antero52
A detailed foreground and a smooth sky is a particularly tricky combination. People often take bracketed series and work from there, often by HDR techniques. Heres the overall situation. Even if the camera has enough dynamic range to keep detail in the dark foreground and bright sky, normal 8-bit displays or printing papers cannot show all that detail. If the tonal range is simply compressed, the resulting image looks flat. HDR software tries to solve this by using local non-linear processing, usually combining the bracket shots to an HDR shot with an extended (32-bit) dynamic range. No output device can display those images, which is why dark areas are lightened and light areas are darkened. In short, HDR processing is geared towards enhancing detail by enhancing local contrast and suppressing overall contrast. But this often ruins smooth skies because the Detail Enhancer looks for detail that actually shouldnt be there. If the sky has a gradient because some areas are lighter than other, the details enhancer amplifies the gradient to something horrible. It is often impossible to find a single set of settings that gives adequate detail in the foreground and a smooth sky. Quite often it is necessary to make two or more renderings and combine them in Photoshop (or Elements). The detailed version (for the foreground) is often the easier one, the question is just how to balance detail and realism. The sky version is often harder because its difficult to tell Photomatix to not look for detail in the sky. For the smooth sky I use one of the following: 1) one of the original shots from the bracket series, 2) a version by Details Enhancer with Smoothing, Microsmoothing and/or Highlight Smoothness set to high values, or 3) with the other two algorithms of Photomatix, namely Tone Compressor or Exposure Fusion. Sometimes, with lots of detail against the sky (eg trees in the photo), its impossible to combine a HDR-processed version and one of the original bracket shots. The reason is that Photomatix makes a combination of the bracket shots and has to align the shots with each other, slightly twisting image geometry. This is why the original shots do not always align with the HDR-processed version. For really complex work its necessary (and actually time-saving) to create and save the interim 32-bit version and work two or three different renderings from it. This way you can ensure that all the component shots have been aligned identically.