Umesh Bhatt

Warning: This post is not about how to take HDR and process them. If you are looking for a basic or advanced primer on HDR processing – I would like to refer you to Christian’s Blochi’s HDR book, Dan Burkholder’s Workshops or Trey Ratcliff’s excellent and free definitive guide for HDR processing.

This post is about what these people dont mention in their books/videos/tutorials and such. This post is about my personal experience in processing HDR images and my own tidbits.

Most people use Photomatix – this is a given. However, Photomatix is no magic tool that makes HDR automatically. What most of these experts dont say is all the other tools and techniques they use besides Photomatix to make their HDRs.

Lets cut to the chase – Some claim not to use anything else actually, and I don’t believe it. Therefore this post.

HDRs are most useful when one is trying to capture an image with a huge dynamic range (DR) – a range beyond most consumer cameras (Most DSLRs have less DR than medium format and those are still low compared to what we see in nature). A typical example for a true HDR is here. The shot captures good detail in both the shadows and highlights, and despite the very bright sunlight outside, captures the inside of the house very well. So are the people who are actually on the shade.

So, how exactly do you create one? Well – for starters, there is the typical multi-exposure capture. Most DSLRs can capture about 5-7 EV (a.k.a “stops”) DR. Extending this range is just a matter of over/under exposures. Many modern DSLRs can automatically do this – using a camera feature called “Auto Bracketing”. D300/D3 can Auto bracket +/- 4EV (9 stops: 1 correct, 4 over and 4 under) – This is sufficient to extend the DR to at least 12 EV or more (equivalent to that of medium format camera sensors).

Most experts agree that one must use a Tripod when doing this Auto Bracketing (Manual bracketing is sometimes recommended if the DR of the scene is very high such as a very dark room, with a window view in bright noon sunlight). Shooting or using a tripod is actually not possible in certain situations – for example, you are shooting in a place that prohibits tripods, or shooting images in a vacation (I recommend not to carry tripods on vacations). The good news is that I have successfully shot many HDR images with good success hand held. The trick to doing this is to have a very steady hand, and use the fastest continuous mode available in the camera. This may sometimes still not be sufficient to get good alignment (stacking the multiple exposures).

Here are a few notes from my personal experiences:

• Shoot RAW when possible. The D300 allows me to shoot 12 bit RAW in high speed continuous mode at the rate of 6 fps. Using a vertical grip, gives even a higher speed (8 fps).
• Use the camera manufacturer’s software to import and process RAW images rather than a 3rd party RAW converter. This will give you the *best* possible image to work with. I have found out the wrong way that doing otherwise leads to white balance issues that takes hours to correct after the HDR is processed. I typically use Capture NX in batch mode, to process the RAW images (imported by Lightroom) and convert them to TIFF files.
• If the source images are not noise free, try running Noise Ninja or equivalent to remove noise at this time. If you have used a very wide angle lens to capture your source images, use PT Lens or equivalent to remove the distortion before using your images for HDR conversion.
• If your favorite HDR processor does not do perfect alignment (View HDR file using the HDR viewer at 100% to determine if the edges are perfectly aligned). Do not continue to the next step (Tone mapping) if the HDR image is not perfectly aligned. The resulting images are simply not sharp enough to print. Instead, import the images into CS3 using the option “Scripts->Load Files into stack” option and select “Auto align” to align the layers. Once aligned, you can use the option “Scripts->Export layers to files” to write them back as aligned TIFF files. This will give you perfectly aligned source images for HDR conversion.
• For HDR Panaromas, i recommend that one create the series of HDR images first and then stitch them together. You have the option to do this automatically if you use a tool such as PT Gui Pro, that lets you just dump in all source images, and it blends and stitches it at the same time.
• For HDR image conversions, I typically use the Batch command option to process each HDR (Each series multi exposed images are in its own folder) to blend the images into .hdr files. The tone mapping is always done manually.
• For tone mapping, i almost always do it conservatively. I typically use “High” to “Very High” light smoothing setting, with the strength around 90 (If you see halos, start reducing this number until it goes away), color saturation about 70, color temperature about -3 to -5 (very important, otherwise most images will have an orange cast), highlight/shadow smoothing at 1 or 2 and Gamma at 1.2.
• The other other setting i usually vary for each image are the black and white points.
• On occasions, i repeat the tone mapping for 2nd time to get even more pronounced effect. When i do this – i often save it as a separate file, and merge the first tone mapped image into this with masks to show only those parts that need extra details.
• In those situations that have people in the HDR shots, i have two options – to clone out the people or to blend the original correctly exposed image on to the HDR shot with masks to enhance regions from HDR on to the correctly exposed image.
• I have found that on occasions, i will need to correct for color casts despite all the other measures. You can easily do this in Photoshop, by duplicating the merged layer, run “Filter->Average”. Then create an curves adjustment layer and select the gray dropper and click anywhere on the image. Your image has now been corrected for color casts. Just remove the gray layer and you are all set.
• You may optionally (i typically do this) add more curve adjustments, sharpening (Clarity type of sharpening) and Saturation (Vibrancy type of saturation). You can optionally adjust black levels as well to make the image pop.

That is it. Your final image should now be ready for display.