Open Source, Easy "HDR"
Posted 24 September 2011 - 10:06 PM
Posted 03 October 2011 - 02:50 PM
Posted 04 October 2011 - 08:31 AM
With a HDR (High Dynamic Range) image, you take multiple pictures of the same scene with different exposure settings on the camera (perhaps at -3, -2, -1, 0, +1, +2, and +3 f/stops) so some pictures are over-exposed and some are under-exposed and combine them so you can see detail in both the shadows and the bright areas.
Posted 28 November 2011 - 07:29 PM
Tom's got the gist of it. An HDR image is actually an image with more than the normal range of brightness. A "normal" photograph, for instance, comprises roughly 5 or 6 stops of brightness. Put another way, the brightest part of the image might be roughly 50 times brighter than the darkest portions (one stop=twice as bright, 2=4 times as bright, 3=8 times, etc.).
Monitors and printed images suffer from the same limitation. True HDR images however use more than the standard 8 bits per color to represent brightness, so they must be converted to 8-bit per color for display or print (the converted file is referred to as an LDR image, which is actually just the normal 8-bit per color image type we're commonly used to).
The trick with the conversion process is to present an image that shows the contrast over the entire brightness range without altering things to the point of looking unnatural. There are many ways to do this, but none seem to work in all cases. Many methods result in rather extreme alteration of local color and contrast.
The method I described doesn't alter any of the data in the images (it's not true HDR, since there is never really any computation of higher dynamic range), but rather combines the data from bracketed exposures to represent a wider dynamic range without tone mapping. Tone mapping most often alters local contrast and color to the point where the image takes on an unnatural look and introduces artifacts such as dark halos in high contrast areas of the image.
This method avoids that, at the expense of not being able to adjust parameters. It's not a bad trade-off, since the parameters you'd be adjusting don't really apply. They are mostly concerned with minimizing tone mapping artifacts which won't be created at all using this method.
I find myself using this method over any of the HDR software I've used since it doesn't play havoc with colors or produce "obvious HDR" images.