Kevin Childress Photography | Maximizing Your HDR Photography

Maximizing Your HDR Photography

April 26, 2015  •  Leave a Comment

In the world of photography, one can take several different paths to reach similar end results, and nothing could be more true than with HDR (High Dynamic Range) photography. And in the world according to me there are good, better, and best techniques for producing those HDR images. This article is written to discuss what I believe is the best approach using techniques I have developed over the last several years with lots of practice along the way. If your goal is produce gritty, grimy, techno-crap images, then you can stop here. If your goal is to to produce technically superior HDR images, read o​n!

St. Luke's Episcopal Church in Lincolnton, NCFig. A: St. Luke's Episcopal ChurchNikon D800 - Sigma 10-20mm at 10mm - ISO100 - f/16 - 15 exposures at +/- 1.0 EV ranging from 1/320 - 50 seconds Too often I hear people asking what is the best HDR-imaging software program. And while its true that some programs are better than others (we'll get to that later), superior HDR photography begins with precise camera work. And depending on the scene that could be a lot of camera work. In short, my philosophy is that one must capture the entire dynamic range of a scene in-camera in order to collect the necessary data that is needed later during the tone blending process. That means that if a scene's highlights meter at 1/320 second, and the shadows meter at 90 seconds, then so be it. That's why we call it high dynamic range photography and you will be well served to capture each and every stop of light between 1/320 and 90 seconds. Notice that I said every stop of light, not every 2nd, 3rd, or 4th stop. In the example of 1/320 to 90 seconds, let's call that 16 stops of dynamic range, and you need all of them!  My approach has become to capture those 16 stops of dynamic range in 1-stop increments. Figure A above is a good example where 15 exposures were used for tone blending. 

So what's the difference in capturing 13, 14, or 15 exposures at +/- 1.0 EV instead of 3 to 5 exposures at +/- 2.0 EV? What's to gain you ask? The answer is simple: more data. And more data equals greater fidelity. And when it comes to post processing (which is obviously a must in HDR photography), your tone-blending software will make much better decisions with more data. A key part in maintaining excellent image quality in post processing is the transition between highlights and shadows and keeping luminance and color noise to an absolute minimum. Consider this: for every stop of light that you don’t capture in-camera, you are relying on your tone-blending software to interpolate the missing luminance and color data for filling in the gaps between widely varying exposure values. That interpolation is where much noise and posterization is generated, and the result is degraded image quality. We could probably all agree the more data a computer has to make decisions, the more accurate the computer’s decisions will be; the same goes for digital images and that philosophy applies doubly when blending color and tones from multiple exposures into a single file.

Duke Chapel in Durham, NCFig. B: Duke ChapelNikon D800 - Sigma 10-20mm at 10mm - ISO100 - f/16 - 10 exposures at +/-1.0 EV ranging between 1/50th to 10 seconds

We’ve seen enough HDR images created in recent years using 3-to-5 exposures at +/- 2.0 EV that we've become accustomed to the results produced by that approach. But let’s face it, the +/- 2.0 EV approach is just not the perfect one-size-fits-all-dynamic-ranges photography tool, particularly for many interior spaces. Most church interiors that I have photographed run around 13 stops of dynamic range. While the +/- 2.0 EV approach is the quickest route, the highlights and shadows typically get stretched way too far during local tonemapping and the midtones are left to bridge the gaps. As I mentioned, that can lead to a lot of posterization which is never a good thing, especially if you're going for excellent image quality. Figure B above of Duke Chapel is the final result of 10 exposures ranging between 1/50th and 10 seconds. Figure C below is a screenshot of the 10 exposures that were used. Looking at the first frame you'll see the brightest highlights (the chandeliers) are barely exposed. And if those chandeliers had shown bare light bulbs I would have exposed them even less - maybe a little as 1/800 second. You need that sort of nearly-black frame to protect the highlights when processing your final tone-blended image. 

Fig. C Many folks rely on adjusting the exposure value of a single raw file for producing “multiple exposures” and then re-blend those files into a faux HDR image. But why do that when you could have captured a few more stops of light to begin with? If you blow the highlights in a single take, they’re gone. There’s simply no detail in blown highlights and no matter what adjustments you make to the original RAW file, you will never regain that lost detail. There is only one solution to this problem: more data, and more natural exposures. Don't waste your time adjusting exposure values in multiple RAW files. For every exposure adjustment you make to those individual RAW files, you're introducing more and more likelihood of luminance noise. Slow down, take your time, and get the files you need from the beginning! Your final project will be much, much cleaner if you put in the necessary work up front.

Fig. D: Lamp Detail Figure D at right is a 1:1 screenshot of an example of the type of detail you can retain if you protect the highlights with the initial photography. Click on the thumbnail to enlarge the image and look at the circular grates on the bottom of the lamp. Those grates are concealing bare light bulbs, and while the bare bulbs aren't visible at this angle, those areas of the lamps are super hot with light. The fact that the grate detail is even visible is because the initial photography included a nearly black frame (probably around 1/500 second) to protect the highlights of the interior lights. 

Fig. E: Christ Episcopal ChurchNikon D7000 - Sigma 10-20mm at 10mm - ISO100 - f/16 - 13 exposures at +/- 1.0EV ranging from 1/50 second to 60 seconds Finally, Figure E at left is tone-blended from 13 exposures. In theory, 4 exposures at +/- 4.0 EV could have produced the same results, so I tried it with the exact same tonemapping settings I used with 13 exposures. From a broad perspective, I did get similar results. The differences are in the details like harsher light falloff directly at the light sources (especially around the windows). The 13-exposure version has a smoother transition to midtones in these areas and the colors on the windows look a tad better. Also in the 4.0 EV version I see the super-light highlights (like the light bulbs) aren't quite as white; they have begun to take on that gray tone that we see on so many HDR images when the highlights are beaten into submission. Finally the 13-stop version gives you far better noise-to-signal ratios and thus produces much cleaner, less noisy images.

Nikon D800 - Nikon 14-24mm f/2.8 - 14mm - f/16 - ISO100 - 9 exposures at +/- 1.0 EV One might think this article was written specifically for interior photography but that's not the case. I regularly use this same technique in landscapes, particularly with waterfalls. I think the biggest challenge we have with HDR photography is movement in the scene between the frames that leads to ghosting in the composite image. You simply have to use your own discretion at the moment of capturing your frames as to how much time you can allow to elapse (which effects the number of frames you capture) depending on whatever motion there is in the scene. The waterfall shown here gives the appearance of motion in the water, which is my preference for photographing waterfalls. This was photographed on a very calm morning with no breeze so there is no apparent motion in the leaves. This waterfall was shot with 9 exposures at +/- 1.0 EV.

Wrapping Up The Photography: The bullets below list my basic workflow for the camera work discussed above. 

  1. ​Set manual white balance. Remember that you're bracketing a bunch of frames so don't rely on auto WB here. The slightest change in light quality can really shift the color pallete across the bracket with auto WB. 
  2. Meter highlights.
  3. Meter shadows. Might require some guess work for exposures that exceed 30 seconds.
  4. Align lens and lock the tripod head tight.
  5. Set focus. If shooting an interior, I typically select a single AF point. If shooting landscapes I typically use the full AF array.
  6. Switch camera/lens to manual focus. This ensures the focus point does not change across your bracketed frames. Be careful not to touch the focus ring on the lens after locking focus.
  7. Capture the frames (and I strongly urge you to use a remote shutter release): My D800 can bracket up to 9 exposures so I typically only have to advance the bracket once when I need more than 9 exposures. My D7000 can only bracket 3 exposures. Clearly the D7000 requires much more camera work to capture all the frames that I want. If your camera can only bracket 3 exposures:
    • Set the camera to capture the exposure bracket progressively.
    • Begin your first bracket with the shutter speed metered for the highlights (the fastest shutter speed), and capture the first three frames.
    • Manually advance the shutter speed to one stop slower than the last frame captured in the previous bracket.
    • Repeat process until you have all the exposures metered between the highlights and shadows.

 

Post Processing: Where do I begin?! Post processing of HDR images is one of the most hotly contested subjects of the last decade in photography. And there is no question that post processing your HDR images can make or break the whole deal. As you may have guessed by reading my opening statements, I am not a fan of the gritty, grungy, techno-crap-tone-mapped images that has given HDR photography such a bad rap. I prefer to process my HDR images to be as realistic as possible - to look as much like a "perfect single exposure" as possible. Of course I have never duplicated what my eyes have seen with an HDR photograph, but I do the best I can based on my memory of the scene and with the impression it left upon me. Today there are many programs available for blending all your exposures into a single file, and I won't even begin to review them all. I will simply comment on the tools I use today and what has proven to suit my style the best.  

Today I use 32-bit floating point TIFFs exclusively for my HDR composites. I still have Nik HDR Efex Pro2 on my PC, as well as several other capable programs, but I wholly believe that 32-bit floating point TIFF files are the holy grail of HDR post processing. My high-level post processing workflow is:

  • RAW file development in Lightroom.
  • Export developed RAW files as 16-bit uncompressed TIFFs in Lightroom.
  • Merge all the uncompressed 16-bit TIFFs into a single 32-bit floating point TIFF.
  • Import new 32-bit TIFF into Lightroom and develop the image just as if it were a single RAW file.
  • Export adjusted 32-bit TIFF into Photoshop CS6 for final rendering with luminosity masks and whatever adjustment layers I deem necessary to meet my vision.

Raw File Development: For this example let's just say I'm working on a bracket of 12 exposures. Since I have 12 one-stop exposures, that is all the luminosity data I need. I NEVER adjust exposure, contrast, highlights, shadows, whites, blacks, hue, saturation, or lightness during RAW development; I want those elements to stay as "natural" as possible in the individual exposures. This is my RAW development workflow:

  1. Choose a single file to work with - usually from the middle of the bracket.
  2. Lens corrections if needed including perspective and color.
  3. Straighten image if needed, but perform no other crop.
  4. Verify white balance. If needed, select a manual WB point from the image. Refer to Figure E above: I used the white in the flags hanging in the pulpit to validate WB in the source RAW files. 
  5. Add a touch of clarity (usually no more than +5 on the Lightroom sliders).
  6. Noise reduction.
  7. Sharpening.
  8. Copy adjustment settings and paste to the other 11 RAW files.
  9. Export all 12 files as 16-bit uncompressed TIFFs.

Merging 16-bit uncompressed TIFFs to a 32-bit floating point TIFF: For the past couple of years I have been using the free trial version of Photomatix for this function. I do not use Photomatix to make any adjustments to the image, I ONLY use photomatix to facilitate creating the 32-bit floating point TIFF. Within Lightroom I have created an export agent that interfaces directly with Photomatix so this helps streamline the process a bit. The process works like this:

  1. Select all 12 TIFF files in Lightroom.
  2. Right click on the selected files > Export > Export to Photomatix PRO.
  3. Once the Photomatix import dialog opens I select the following options:​​
    • Align Images by Matching Features. 
    • Remove Ghosts with Selective Deghosting Tool. And I strongly urge you to select any ghosting areas manually. 
    • Reduce Noise on Underexposed Images Only. 
    • Remove Chromatic Aberrations (even though I've already done this in RAW development, it doesn't hurt to get a second opinion!).
    • Show Intermediary 32-bit HDR Image.
  4. Now Photomatix will import all the 16-bit uncompressed TIFF files and will run through the aforementioned script. 
  5. Once Photomatix creates the HDR image it will display the intermediary 32-bit TIFF. The image will look absolutely horrible but don't worry about this right now.
  6. Select FILE > SAVE AS > save as Floating Point TIFF.
  7. Import the new 32-bit floating point TIFF into Lightroom.

NOTE as of 4/26/15: Last week I installed Lightroom 6 which includes the new Phote Merge to HDR feature which will create the 32-bit floating point TIFF. I've used it a couple times in testing and it looks pretty slick. But I can't comment right now on how well the deghosting feature works so I'll come back to this at a later time ...

Adjusting 32-bit Floating Point TIFF in Lightroom: When you first see this file in Lightroom you will probably think it looks terrible! But you now have a single 32-bit TIFF file that contains a tremendous amount of luminance and color data - just think of this as a super-duper RAW file! But at this point, the "tonemapping process" is to each is own. My suggestion is that you click on the "Auto Tone" button in Lightroom just to get a good starting point and then perform whatever adjustments you feel necessary. I try to keep things simple when I'm processing the 32-bit file in Lightroom. I'll make basic adjustments to setting the white and black points and then exposure, highlights, shadows, and maybe (just maybe) clarity. I'll take a look at color hue, saturation, and lightness but if I make any adjustments here I will typically go easy (save the heavy work for Photoshop). Once I'm done with the 32-bit file in Lightroom: Right click on the file > Edit in > Edit in Adobe Photoshop > Edit a copy with Lightroom adjustments.

Adjustments in Photoshop CS6: This process is very individual based on each of our styles for post processing. I won't try to explain all that I do in CS6 because clearly that can get very laborious. I commonly use several targeted color adjustment layers for hue/saturation/lightness. And I will finish the image with whatever luminosity masks I need for fine-tuning exposure and contrast.

And in a nutshell, that's it!  :)  I would be most appreciative on any feedback you have related to this post.

Wishing you the best of happy photography!


Comments

No comments posted.
Loading...