I'd like to get some recommendations for books on HDR for beginners. I have zero experience in this area, but would like to understand how HDR can improve images beyond what can be done with traditional photography. No, I'm not interested in the grungy, surreal side of HDR, I hate that with a vengeance.
Hoi Frans,
Welcome to the wondrous world of HDR Imaging (HDRI).
I just spent $60 on Christian Bloch's The HDR Handbook 2.0 and I can't say that it was of much help.
That's too bad, because it is a must have read book for the theoretical background, and it also offers a good comparison of the various hardware and software solutions, that were around at the time of publishing of course.
It lacks good examples of before and after for different situations and how to achieve it, so I still don't have a good feel for what I'm after: how exactly does HDR help to do what can't be done traditionally. Any suggestions?
Maybe your disappointment has something to do with the expectations? HDR imaging is not that magic bullet that some believe, but it is an essential technique to capture the wide dynamic range in some scenes, and allow post-processing (tonemapping) without creating noise issues or featureless highlights and shadows.
The HDR part of HDRI is nothing more than
capturing the wide dynamic range of the scene with a device that cannot capture that range in one exposure. Even if it can, or almost can, the captured shadow areas will have received relatively little information in the form of photons. That will mean that the shadows will be relatively noisy, which only becomes more noticeable when you start post-processing them to e.g. make the shadow details a bit more visible.
The tonemapping part of HDRI is nothing more than
post-processing the captured image data, much like with normal images. However, the shadow details will have much lower noise, and the highlights may hold much more detail than in a regular image. If one were to squeeze the full dynamic range into the limited output range, it would look very low contrast and lifeless. Afterall, what was blindingly bright or deep black in the original scene, will now be reduced to e.g. a contrast ratio of 100:1 on paper output. It will take a lot of contrast manipulation to fool our eyes into believing that we are looking at a high contrast scene, and because we now have better (low noise) source data, we can indeed push the global and local brightnesses around much more to create that natural looking impression of the original scene.
Maybe that is also why it's difficult to demonstrate before and after effects in natural looking HDRI. The before situation only exists at the time of shooting, and only you know that. The scene lighting may have been one of the reasons why the image was shot in the first place. After that, you are left with an abstraction of that initial impression. It now depends on which aspect you want to stress or suppress because it might distract in the composition.
Currently I'm still most pleased with the tonemapping of
SNS-HDR, which allows to create very natural looking renderings of difficult scenes. It responds real-time as the sliders/controls are moved, so one can focus on the image preview while changing the settings and immediately see when to stop or when to go a bit further (assuming one has a goal in mind). It's very much a creative process, almost like painting with light, but with a huge amount of control.
Having technically better (low noise) image data also allows better use of other useful post-processing options, such as
Topaz Labs Clarity, which already performs miracles on LDR images but now has much more low-noise information to work with.
So, I suggest you make the clear distinction between the capture part of HDRI, and the post-processing part of HDRI. Both are needed for a successful result, and both have their issues to overcome. There are tools to assist in that process, but it also requires a clear goal towards which to work. It also still requires good composition and an interesting subject to make beautiful images, but the scene dynamic range should be less of an obstacle to pull it off in the end. As always, there will be a learning curve, so practice will help.
Cheers,
Bart