Update 1/19/12: Based on feedback from many including John Omvik at Unified Color, I’ve improved my #1 workflow and substantially edited all the workflow descriptions below.

This is my first-ever high-dynamic-range (HDR) image, shot nearly three years ago. Since then my HDR workflow has changed quite a bit — almost weekly, it seems. As I’ve recently been running tests on some aspects of HDR processes and tools and particularly methods for transferring images between applications, I thought this would be a good time to pin down and document the workflows I’ve been using and explain why I’m still using them or not.

My primary applications are Adobe’s Lightroom (LR) and Photoshop (PS). I use LR to organize my images and for basic processing. I turn to PS for images that require adjustments beyond LR’s capabilities. The HDR tools I’ve used at one time or another include Photomatix Pro, LR/Enfuse, HDR Efex Pro, HDR Expose 2 and its cousin, 32 Float. Note that all of these applications are available as free-trial versions. I encourage you to download them and experiment with these workflows.

Because you may not have some of these applications, I’ll describe seven different multi-image workflows as well as some for single-image HDR. My goal is to cover not only the tools and methods I’m currently using, but also those that I tried, tested and in some cases abandoned before settling on my current choices.

Workflow #1: 32 Float (with Photoshop)

I’ve recently started using Unified Color’s 32 Float for most of my high-quality HDR images. 32 Float is a PS-plugin version of HDR Expose. Given that I usually end up in PS anyway, this gives me a simpler workflow than I’d get with HDR Expose. I’m only giving up batch operations and a few functions that are better performed in LR or PS anyway. This is my current workflow for my highest-quality HDR images.

1. Unless you’re shooting action/sports, which rules out multi-exposure HDR anyway, always shoot in RAW. Otherwise none of this applies to you.
2. In LR, make only two adjustments to your images: lens correction and camera calibration, which can only be done at this stage. Defer everything else until your RAW files have been merged into a single 32-bit image. (I have camera profiles for each of my body/lens combinations made using a ColorChecker Passport.) Use Copy/Sync to apply the same adjustments to all the bracketed originals. If you’re not an LR user, you can instead perform the equivalent of this and the following step using Adobe Bridge and Adobe Camera Raw (ACR).
3. Select the bracketed RAW files in LR then (from the Photo menu or ctrl/right-click) Edit in…Merge to HDR Pro in Photoshop. Importantly, and unlike some other Edit in… and Export options, Merge to HDR Pro sends full-range 32-bit files to PS, even if you’ve made adjustments in LR. This launches PS’s own HDR Pro module. [PS’s tonemapping is weak, which is why all these third-party tools exist. But the merge-to-HDR function seems to work as well as any other.] In Merge to HDR Pro don’t be alarmed if you don’t see your highlight or shadow details. You’re only looking at a low-dynamic-range (LDR) preview of the HDR image, which can’t be properly displayed on your monitor.
4. In HDR Pro set the Mode to 32 Bit and click OK. This merges the originals to create a 32-bit HDR image, then opens it as the Background layer in PS.
5. Launch 32 Float from PS by selecting Filter…Unified Color…32 Float.
6. In 32 Float tonemap the HDR image to low-dynamic-range (LDR). Many operations are best done here, while you’re in 32-bit mode and your luminance data are separate from your color data. Go ahead with global changes such as sharpening, noise reduction, local contrast and color and tone adjustments, but I’d stay away from dodging and burning in particular. At least in the version I have, there’s no undo feature, so an erroneous burn can wipe out what you’ve done, just like in a wet darkroom!
7. You’ve now done pretty much all you can do in 32-bit mode, so set Upon Apply Convert To to 16 bpc. This will return a 16-bit image back to PS and change PS to 16-bit RGB mode. It’s also better to change to 16-bit mode while you’re still in 32 Float since there can be some rather quirky artifacts when the 32-to-16 bit conversion is done within PS.
8. Back in PS, decide whether you want to merge in any of the original exposures. I will often do this in cases where there are ghosts such as people in different positions or if there are artifacts, details, tones or colors that are much better in one of the original images than in the merged one. If so, go back to LR, select the original RAW images and click on Photo…Edit In…Open as Layers in Photoshop… Back in PS, select the merged image then use Layer…Duplicate Layer to copy it as the top layer in the RAW-image stack. Use layer masking and other tools to manually combine your  originals and the merged image.
9. Further tweak the combined image as necessary. If you need a filter that’s only available in 8-bit RGB mode such as Pixel Bender, Distort or the Topaz Labs suite, change the Mode to 8-bit RGB, but do so as late as possible. Otherwise keep the image in 16-bit mode.
10. Click Save, which returns the image to LR.
11. In LR do your cropping, final sharpening, noise reduction and vignetting.

Workflow #2: HDR Expose (without Photoshop)

This variation is designed for those who either don’t have PS or just prefer to do all their retouching in LR.

This plugin is sold separately by HDRsoft. It’s the best way to use Photomatix Pro with PS.

This workflow bypasses PS altogether, moving images from LR directly to/from Photomatix Pro. This process is fine for some HDR images, but it has certain weaknesses. First, it depends on Photomatix Pro to perform the HDR merge and image alignment — functions performed better by PS.

Nik Software’s HDR Efex Pro is the new kid on the block. While I like Nik’s control-point UI and have settled on their Silver Efex Pro as the #1 plugin for monochrome, HDR Efex Pro has become a tool I don’t use much any more. But many people use it, so I’ve included it in my workflows.

When your source is just a single RAW file, there’s no need for the Merge to HDR Pro step. In fact, you can’t run Merge to HDR Pro with just one image. In this case you need to get your RAW file to your tonemapping utility as directly as possible. Starting in LR, use one of these tools, ranked in order of my personal preference:

• LR/Enfuse: Because it’s simple, this is a workflow I sometimes use for single-image HDR. Just follow the same steps described for Workflow #3.
• Photomatix Pro: You can’t send a single RAW image through PS to the Photomatix Tone Mapping Plugin, but you can export a single RAW image directly to Photomatix Pro, which automatically returns the resulting image to LR. Use File…Export…Photomatix Pro, but make sure you change the File Settings in the Export One File dialog box to DNG as described in Workflow #6 for multiple source images using Photomatix Pro.
• HDR Expose: This workflow is more cumbersome than using LR/Enfuse or Photomatix Pro, but it’s the process I use for single-image HDR if I’m not satisfied with the results from the simpler tools. HDR Expose is the standalone version of 32 Float, and it can open one or more RAW files directly. Unfortunately it can’t deal with a single image exported from LR, so you need to start HDR Expose and open the file from there. This means you’ll have to save the output of HDR Expose as a file (typically a 16-bit TIFF) then import that back into LR.
• 32 Float: Because PS’s Merge to HDR Pro won’t accept a single RAW image, the only way to get the full dynamic range of a single RAW image into PS is via the extended-EV TIFF method, described below.
• HDR Efex Pro: This application does not include a RAW file processor. The only way I know to preserve the full dynamic range of a single RAW file for HDR Efex Pro is to use the extended-EV TIFF method.

This is a method for recovering data from RAW files when the tools you’re using cannot read those RAW files directly. It is not required or recommended for any of the multi-image workflows above. The only time I use it is when I want to process a single RAW file using 32 Float or HDR Efex Pro, which is now pretty much never. You’re probably better off just using another tool such as LR/Enfuse, Photomatix Pro or  HDR Expose, but if you don’t have one of those applications or plugins, the following is your best choice.

I first learned this technique from Klaus Herrmann in the section in his online HDR Cookbook entitled Creating HDR Images the Right Way. (Look for the Five TIFFs method.) A RAW file can contain image data from the darkest shadow detail to the brightest highlights, spanning a range of 10EV-12EV. But an LDR file such as a 16-bit TIFF can only represent 6EV-8EV. The idea of the extended-EV TIFF method is to replicate the wide dynamic range of data found in a RAW/HDR image using a bracketed set of LDR TIFFs from that image. Each TIFF file will contain the data from a different (but overlapping) portion of the RAW image’s brightness range.

1. Open the RAW image in a RAW file processor application such as Lightroom, Adobe Camera Raw (Photoshop), Phase One’s Capture One Pro, Nikon’s Capture NX2, etc.
2. Using the app’s Exposure slider or equivalent, reduce the luminosity by 4EV.
3. Save the image as a 16-bit TIFF, preferably in the ProPhoto RGB colorspace, with a filename that both identifies the original image as well as the adjusted EV. Something like “IMG1234_-4ev”.
4. Repeat the previous two steps so you end up with five TIFFs, one each with exposure adjustments of -4EV, -2EV, 0EV, +2EV and +4EV.
5. Check the +2EV and +4EV images and decide if they should be included in the set. While the -4EV and -2EV are very likely to include extra highlight data that appears blown out in the 0EV image, the same is often not true for shadow recovery. If the +2EV and +4EV images don’t contain true shadow detail that doesn’t appear in the next-lower-EV image, don’t use them — they won’t add detail to the shadows, but they will increase the noise.
6. Treat these three, four or five images as bracketed originals and submit them as input to your HDR merge application such as HDR Pro (Photoshop), HDR Expose, Photomatix Pro and HDR Efex Pro. You can see the results of my tests of this method at HDR Tools Comparison.

If you’re working with multiple bracketed RAW originals and you’re still getting blown-out highlights when tonemapping, you may be able to recover them using a variation of the extended-EV TIFF method:

1. Create a 16-bit ProPhoto RGB TIFF with no exposure adjustment from each RAW original.
2. From the darkest (lowest-EV) RAW image, create two additional TIFF files: one darkened an extra -2EV and another at -4EV.
3. If you started with three RAW images, for example, you’ll now have five TIFFs.
4. Use these TIFFs instead of the RAW files as the source images for your HDR merge process.
5. As with the single-RAW image variation, you can also try +2EV and +4EV TIFFs made from the brightest RAW image, but again to avoid noise don’t use them unless they provide details in the shadows that don’t already appear in a lower-EV image.

[Please post all comments questions on this Google+ page since there are far more photographers there than are reading this blog.]