• Andy’s E-BOOK — Photography Travel Guides


    All Images and writing on this blog are copyrighted by Andy Richards. All rights are reserved. You may not, without my express, written permission, download, right click, or otherwise copy my images for any reason. Copying an image and putting it on your blog, website, or even as a screensaver on your computer is a breach of copyright, EVEN IF YOU ATTRIBUTE THE SOURCE! Please do not do so.
  • On This Blog:

  • Categories

  • Andy’s Photography Galleries

    Click Here To See My Gallery of Photographic Images

    LightCentric Photography

  • Andy's Flickr Photos

  • Prior Posts

  • Posts By Date

    September 2010
    M T W T F S S
    « Aug   Oct »

Managing Dynamic Range Digitally (a comparison of HDR conversion methods and software)

Sante Fe Ski Basin - HDR conversion from 5 separate exposures

In March, 2009, I blogged about “High Dynamic Range” photography.   I had newly purchased the Photomatix program, after researching the available alternatives.  Since then, Adobe has released CS5 and with it their own HDR rendering program, “Merge to HDR pro.”  I thought it was time to give more detailed treatment to this technique.

For those whose photographic experience dates back to film, we remember how important the choice of film and the light conditions were.  It was only those who had the luxury of their own color darkrooms, or a very good (and expensive) custom lab, who could so more than try to capture the most information they could on the particular film emulsion.

Film was said to have “exposure latitude,” which was measured by f-stops or the more modern name: “exposure value” (EV).  Color slide film, which was probably the most commonly used film by pros and by advanced amateurs, had, at best 3 stops of latitude.

Exposure lattitude is now referred to in the world of digital capture, as “Dynamic Range.”  Digital photography, particularly when shooting raw images, has revolutionized dynamic range, perhaps as much as any other facet of the technology.  Today’s digital SLR sensors are able to capture as much as 5-7 stops with a single raw capture.  This means that with careful attention to exposure, and with careful editing during the raw conversion and post-processing, we can get a lot out of our single captures.  But that is only scratching the proverbial surface.

The real story is the ability of digitally captured images to be blended and combined.  We are now capable of displaying digitally combined images that are much closer to what our eye can see (which is significantly more than film or single capture imaging devices can – so far.  I expect to see HDR sensors available well before I leave this earth).  Currently, with post-processing software, we are really only limited by the abilities of our display media.

What this means to us as digital photographers is that if we have a “still” subject with relatively stable conditions, we can – with a series of shots — capture raw images that preserve virtually every important exposure detail in a photograph, and combine them in a way that allows us to finally display a single image with all those important details.  This should make difficult lighting conditions easier to handle, and gives us one more approach, to keep in our arsenal.

I would not in any way consider myself a “graphic artist.”  These techniques open the door to some pretty creative things that can be done with an original image.  Indeed, “HDR” for many photographers, has come to mean an almost surrealistic depiction of an image which looks unreal.  That is not my purpose either personally, or here, in describing the use of these programs.  I am striving, within my own tastes and views, of course, for a photorealistic image.  These programs are well capable of doing that, if thoughtfully and carefully used (more on that later).

Stone House using Photomatix "tone compressor" conversion

I recently captured images in Manassass Battlefield National Park of an old stone house that was used as a Union field hospital during the battles of Bull Run.  Using 3 images each 2 stops apart, I used Photomatix’s two different converters and the new “Merge to HDR pro” built into Photoshop CS5, to compare them.  I ultimately concluded that Photomatix still holds an edge over CS5, but they are close, and it may very well vary from photograph to photograph.

There are a number of steps that need to be taken to prepare images properly for HDR.

  • The subject must be still. Moving subjects and subjects with movement within them (for example, a coastal scene with waves) are problematic.  You may be able to set your camera’s bracketing feature and capture images with a very fast shutter speed in some instances.  However, many landscape scenes will lend themselves well to HDR.  The software does have some pretty sophisticated algorithms for removing some ghosting (i.e., blowing leaves), but this will be an issue.
  • Since we are essentially combining layers, in most instances, we want to have identical focus and DOF.  Therefore, you need to set your camera aperture constant, and vary the exposure with shutter speed (again, a reason why movement will be a problem).
  • The experts suggest that when using raw capture, you should use at least a 2 stop EV differential. Any less than that will probably result in overkill, and any more will risk losing important information.  Like any other aspect of photography, these are general rules and you may find good reason to vary them depending on your subject, conditions and your personal experience.  Don’t be afraid to experiment.
  • Before bringing the raw images into the HDR processing program, open them in the raw converter and do some basic post-processing.  I use ACR.  I opened these images, sharpened them (I will have an upcoming post on the new sharpening algorithms in ACR 6.0), and synchronized the sharpening and white balance.
  • After doing the HDR combination, I have learned (the hard way) that it makes sense to save the 32 bit resulting file (there are several formats; I chose the .exr format).  The files are very large, but it saves a lot of time going back and forth.

In my March 2009 blog, I recommended, and still recommend, two very good books on HDR, “The HDRI Handbook, by Christian Bloch, and “Complete Guide To High Dynamic Range, by Ferrell McCollough.  The authors go into great detail regarding the theory of HDR, and how to capture images destined for an HDR program.  They also cover the various available programs.  The most popular among photographers continues to be Photomatix, largely, I believe, because of its easy-to-use interface (I tried FDR Tools, partly because I read that it yielded more “photorealistic” results.  But I found the interface very difficult).  NIK software has recently announce the release of its HDR pro software.  I have not used any of the NIK software products, but have seen results and heard from others who swear by them.  I would expect HDR pro to be a contender in this relatively new marketplace.

Essentially an HDR program combines the pixel information from the images into one, HDR file.  The complete dynamic range of this file cannot be displayed on any technology currently available, and the HDR viewers supplied show a combined image that looks rather garish.  In order to make a final image that can be viewed on a monitor and perhaps more importantly, printed, a second process must be done, known as “tone mapping.”  This process takes the important information and “compresses” it into a viewable “Low Dynamic Range” (LDR), or perhaps more accurately, a normal dynamic range image.

Stone House using Photomatix "details enhancer" conversion

Folks who are an awful lot smarter than I am created the software that does this.  There are essentially two different kinds of tone-mapping processors.  One type is “global” in nature and one allows “local mapping adjustments.”  In Photomatix, the “tone compressor” method is global, and the “details enhancer” is local.  The difference, as you might think, is the degree of individual control over the process.  However, the tradeoff is that the local method is much trickier, and can have some very destructive tendencies, unless gently and subtly used.  The details enhancer has a very strong tendency to create artifacts during the conversion process.  The two most common are severe halos around areas of contrast and unpleasant color shifts, particularly in certain areas, such as skies and water.

Note, in the image below “illustrating” these artifacts, how using too much compression in the “details enhancer” crushes the blues and cyans, pushing the sky to near black and creates an ugly halo around the lone tree.  Even in the final version above (in which I significantly reduced the “strength” slider) there is a noticeable halo.  I took it into Photoshop and used a mask to try to eliminate it but wasn’t as successful as I would like (my good friend Al Utzig was able to show me a much better job, it still wasn’t completely eliminated).

Stone House Photomatix "details enhancer" illustrating artifacts

In contrast, note the complete absence of the halo in the tone compression version above.  The tone compressor does not produce any of these artifacts, but you have much less ability to fine tune local areas within the image during the process.  In this case, I like the tone compressor conversion better than the details enhancer or the Merge to HDR versions.  Perhaps a more skilled user could make a much nicer image out of the details enhancer version.  But to an extent, that illustrates a point.  The tone compressor yielded an almost “out of the box” nicer, smoother, more photorealistic image.

Photoshop’s “Merge to HDR pro” has more local adjustments to it, but is much more “fiddly” in my view.  And, I did not see that the results were as “dramatic” in the cases in which I used it to combine images.  My conclusion (and I believe it is joined in by several of the experts out there) is that Adobe still has a way to go to get it “just right.”  In Adobe’s defense, these programs are 32 bit and until CS5, they really couldn’t compete on an even playing field without a stand-alone program.  The CS5 add-in still took, as I understand it, some significant re-programming.  Note how the sky in the left portion of the photo has essentially lost any detail.  Again, perhaps with some experience and greater skill, some of that could be improved.  It may also be possible to pull in a layer masked sky from one of the versions, letting the HDR, tone-mapping process deal with the foreground.  Again, the point is, I want an HDR program that is going to give me a pretty darn good, “out of the box” result.

Stone House using PS CS5 "Merge to HDR pro"

The Santa Fe Ski basin shot is made using the Photomatix “details enhancer.”  On this image, I tried the tone compressor conversion and didn’t care for the result.  In both cases, the Photomatix tone mapping process resulted in a color balance issue, which I found easier to fix in Photoshop.  Again, experience and skill in the Photomatix program may allow me to work with the color issues during the HDR to LDR conversion.

My conclusion at this point is that I will do most of my HDR work in Photomatix, saving the 32 bit .exr image, and trying both conversion methods to see which one yields the most pleasing result.