I have been developing and improving my HDR workflow for months, and would like to explain briefly (edit: not so briefly!) how I make these photographs. Some of the techniques have been adapted from tips in books or online, other steps are rather more trial and error. I am by no means an expert, in either taking good photos or processing them, but I am always keen to learn and improve my technique. I apologise if the steps below sound complicated, or if you fall asleep before making it to the end

Perhaps I shouldn't say too much here, but I don't think anything I do is out of the ordinary. It is a combination of what I have read from a wide variety of sources, and some experimentation. It isn't quite the entire process, but covers most of it.

Pre-Processing

1. Most importantly, try to take a good photograph. This generally means following the rules of composition, and finding an interesting subject in the first place. As I said in my first HDR article, no amount of post-processing compensates for taking a good photo. The HDR technique seems to work best outside on sunny days, although I believe it will also enhance "dull" photographs too. It also works extremely well on architecture. Personally I think the contrast of light and shade is a big aspect of making a photograph appealing. This is known as Chiaroscuro (more on that later).



2. The enemy of HDR is movement. So, on a windy day, take photos of buildings rather than trees! There are various tricks to get around the movement problem when post-processing, but they are time-consuming, and it's better to try picking a subject to minimise the effort. E.g. re-compose a shot so that moving branch is not in the foreground.
   
   Water ripples are another potential hazard. In this case, using a long exposure to blur the water may help. When shooting, it is better to err on the side of under-exposing, rather than over-exposing, as detail can be recovered more easily from an under-exposure.



3. I usually take three bracketed exposures at -2 EV, 0 EV, +2 EV. I use my camera Raw Convertor (Capture NX) for some basic pre-processing; remove chromatic aberrations, and saving off as TIFFs. I also generate some interim exposures, by tweaking the Exposure Compensation. Generally I will increase the under- and normal exposures by +1 to create additional frames. I think this produces a slightly better result. Sometimes I will increase the over-exposed frame by 1 or decrease the under-exposed frame by 1, where clipping occurs. A change of +2 or -2 may yield muddy results, especially when reducing the exposure on an over-exposed area.



4. Noise reduction often improves the results, as noise can be worsened during the HDR merge. I found NeatImage to be the best program to do this. Also, don't necessarily push the settings to the max to remove all noise, as this can often take out the texture with it. Try to get a balance where the noise is reduced, but the texture is preserved. I think the underexposed shots can take some more noise reduction than the normal, or over-exposed shots.



5. Reduce chromatic aberrations. I reduce these in the "raw" files before merging. These are usually purple or cyan fringes which often appear against the sky, like branches or rooftops. Adobe Camera Raw does a good job of reducing these. I also found PaintShopPro's Aberration reduction to be very effective. These can also be reduced the hard way, by making Hue/Saturation adjustments on cyans or magentas, then layer-masking the affected areas.



6. Photomatix. I have tried various HDR generating software packages, but have found Photomatix to give the best results. Photomatix isn't perfect, and the interface could be better, but it seems to be adequate for the job.
   
   When merging the photos, I have "Reduce chromatic aberrations" checked (this seems very effective), also "Reduce noise" and "Reduce ghosting". Reduce ghosting by "Moving objects/people" on High, seems most effective where there is movement, e.g. drifting clouds. However often movement artefacts will need cleaned up later (see below).
   
   Generally, I use the following settings: Strength (100), Color Saturation (60), Luminosity (4-6), Microcontrast (4-5), Microsmoothing (12)... and - most importantly - Light Smoothing (Very High). These seem to produce a decent result for me. I use the Histogram to set the White Point and Black Point to make the curve extend to both the edges, but not bunch up (clip) at either end. This means that the shadows will be black and the highlights white, over the biggest possible range. I usually set the Gamma slightly to the left of 1, e.g. 1.10.
   
   These values will vary from photo to photo, but I find are a good starting point. I don't use the "Smoothing Settings" myself, but I would suggest being very gentle with them.

Post Processing

1. In Photoshop... Auto Levels, Auto Contrast and Auto Color. I will usually apply these and they can make a noticeable difference.


2. Curves, use the eye droppers to select the black and white points to the darkest and brightest areas, also the grey dropper can be used to select a neutral grey area to correct a colour cast. The grey dropper when clicked on a cool colour will warm the shot, and vice versa.


3. Often on landscapes, I'll use a Photo Filter - the Warming Filter at 25% (and Preserve Luminosity).


4. I add some "clarity" to the shot with Unsharp Mask, 25%, 90 pixels. This increases the contrast in a subtle way without really sharpening.


5. Shadow/Highlight to regain some detail in the shadows and highlights. Be very gentle. Around 10%, 10%, 50px for both Shadows and Highlights. Go too far and the result will be awful!


6. If the HDR effect looks a bit too strong, I will blend in the best exposed raw shot on a new layer, Pin Light, at a low amount, like 20%-30%.


7. Next, I change to "Lab Color" mode, and perform some tricks to improve the colour. E.g. adjusting the curves - try steepening the A and B curves, making sure they still cross the centre point. Lab allows you to change the Luminosity (brightness) without affecting the actual colour. Some extraordinary effects are possible by manipulating the curves in Lab mode. Again, I would recommend being gentle!


8. Sharpening. I use the Unsharp Mask in the Luminosity channel. This time, use a very small Radius and a large Amount.


9. Another way of sharpening is to duplicate the layer, use a High Pass filter with a very small Radius. Set the blending to e.g. Hard Light, Soft Light, Linear Light or Pin Light, then reduce the opacity to around 30%.

A few additional tips:

Movement

In the case where there is some movement between in the frames (this may not be obvious until later), I will use the best exposed frame and generate additional frames -2,-1, +1 ,+2. These frames can later be merged to create a photo free of any movement or shifts, however the results are, at best, inferior, and at worst, unsatisfactory. Sometimes I will subtly blend in parts of this "single frame HDR", using a layer mask, to cover parts of the main HDR photo which have been spoiled by movement.

Add Grain

With a large area of sky, there may be some visible "banding", add a little Grain (Filter > Texture > Grain) at low intensity. This is also useful if the photo looks too clean, but adding too much can reduce overall contrast.

Blend in a Black and White layer

Create another tonemapped image, perhaps tweak the gamma a little or the Black and White points. Merge this into the main photo using a new layer, and e.g. Overlay 20%. This can also help bring back some contrast.

Chiaroscuro

I remember this term from Art lessons at school. It refers to contrasts - the word comes from Italian for light-dark. You need contrast to bring out form, solidity and texture. The HDR technique, at first, seems to work against this, by bringing back details into the under- and over-exposed parts of the picture, thereby reducing the contrast. Indeed some badly executed Tonemapping can reduce or even invert the contrast.

For me, this means being gentle with the tonemapping - ensure that shadows remain dark and highlights bright, while retaining some detail, but at the same time, looking for a pleasing balance of contrast over the whole picture.

Some people think HDR means you should pull out every single detail possible over the range of the photograph, just because you can. This is one way to guarantee the "HDR Look" on a photo which many people dislike. As with everything else, I think the key is moderation. Be prepared to experiment and bend the rules a bit.

Hopefully, I have given some of you some ideas which might make a difference - if they work for you, great... if not, then go with whatever works for you. If anyone has any suggestions or advice, please feel free to reply. Thanks for reading!