HDR Workflow Tutorial

I’ll be using the following image of Shenlong waterfall in Yunnan for a step by step tutorial in my particular method of processing HDR images.

The Final result of an HDR image

HDR images tend to be very popular because of its innate properties as well as the fact that the web seems to be flooded with HDR imagery out there these days.  As a personal opinion, I use HDR to present exactly what it stands for : High Dynamic Range in an image – meaning that there are minimal blown highlights and minimal areas of pure black shadow. However, HDR as it has come to be recognised online, has almost become synonymous with the imagery that one particular software program called Photomatix produces. The results are often stunning, surreal, heavily oversaturated and tend to look more like digital art than a photo. There’s nothing wrong with this appearance but I still like to present photographs as photographs rather than digital art as such (definitions clash I know).  I’ve been given the odd compliment here and there that many of my HDR images veer away from the photomatix trend in a good way; which is the reason why I thought I would share some of the workflow and photoshop techniques I use to create the final result. Note that there are many other ways to process and present these images that are by no means less effective or worthwhile.

Step 1: Choosing the right images

One of the important things to recognise when taking shots for an HDR image is that many of the images look awful on their own.  You can take any number of exposures of a given scene and the best results are achieved when you can calculate these exposures manually, however, the easiest way to take an image is to use the camera’s autobracketing function to take three images.

0EV image : neutral exposure : I try to take this image as though I were not shooting with an HDR in mind such that shadows and highlights are balanced as well as possible ( I do use filters even with HDR shots). Often, if I’ve done a decent job, I will only process this image and not go on with my HDR workflow.

Balanced lighting in the image : note the histogram on top right

-2EV image: this image will look dark and unusable but don’t ditch it! The purpose of this image is to have no blown highlights at all so the histogram should be gathered to the left.

Highlights are correctly exposed for (but nothing else) - note the histogram on the top right - this image is probably a little too underexposed even for a -2EV

+2EV image: this image will look blown out as though you’re staring at a lightbulb but once again, this isn’t necessarily a bad thing. Its purpose is to capture shadows in the image exposed to a neutral degree hence the histogram should be gathered to the right.

Ouch! what a sight for sore eyes. Note though that it has served its function by lightening the foresty hills . Note the histogram is bunched to the far right.

The wider the exposure bias in your images, the larger the dynamic range of the final result. If I’m serious about creating a good image, I use a tripod so that all 3 images are exactly aligned. If I’m even more serious, I might take 5 or more exposures for the one scene though I find that in most instances, the standard 3 will suffice.

Step 2:  The Photomatix steps

I wont’ discuss how to use Photomatix at all. Needless to say that it is a very simple process to produce an image from this program. In the screenshot are the typical settings that I use on the tonemapping function of this program.  I choose to use the details enhancer by default.

Importing into photomatix is as simple as dragging and dropping the images into the photmatix window and clicking OK at several prompts

My typical settings on photomatix : others not shown in the other tabs include settings for temperature, smoothing and micro-contrast

The final result from Photomatix’s effort at producing HDR creates some very attractive textures and a heavy degree of saturation. Often, it warms the temperature of the scene somewhat too. It’s shortcomings however are:  Bizarre looking skies that are completely unreal  (a good thing if this is what you’re aiming for) , terrible grain where the program has tried to turn pure blacks into a midtone , uneven lighting which varies from the original image wildly at times.  (In the above example, I was not happy with the darkening of the middle of image,the odd highlights in the water, and the inappropriately darkened bottom right) So this is why I find it necessary to undertake the following steps.

Step 3: Creating the base image to edit:

Typically, I open the Photomatix image as a base layer in photoshop.

I then open the +2EV  image and adjust the colours to match roughly what I want in the final result. Copy the image and add it onto the original image as a new layer and you will see only the +2EV image layer.

The photomatix image is at top left. The +2EV image I've adjusted to match the colour before dragging and dropping into the photomatix image as a new layer

Step 4 : Realistic looking shadowed areas

Create a layer mask over the new layer and select the brush tool. Using black on the layer mask deletes the current layer hence exposing the underlying Photomatix layer. Using white on the layer mask recovers the layer. I use settings on the brush that are soft, and at about 30% strength so that the change is gradual. The aim of adding this layer is often to retain the ‘photographic’ quality of the +2EV image’s shadowed areas while retaining the textures from the Photomatix layer. As a result, I end up ‘erasing’ most of the new layer’s areas where there are highlights in the original scene and retaining most of the new layer’s areas where there were shadows in the original scene.

The two layers visible : Since the +2EV image is on top, that is all you will be able to see at this stage

Setting the brush to small percentages and flow means that any change will be gradual and less likely to need undoing

In the layer mask on bottom right, you can see that the areas I've retained from the +2EV layer are in white. Any areas in black on the layer mask are effectively 'see through' areas to the underlying photomatix base layer

Step 5: Realistic looking highlight areas ( eg water, sky)

Next, open up the -2EV image and adjust the hue/saturation and vibrance to the same degree as you did for the +2EV image.  Copy the image and add it onto the original image as a new layer. You should now only be able to see this new shadowed layer on top of the +2EV layer and the Photomatix base layer.  Create a layer mask as above.  The aim  of this layer is to retain the photographic feel of the bright areas of the original image while getting rid of the shadows from this layer that are unusable.  Often, I end up retaining 100% of the sky from the -2EV image with no contribution from the Photomatix layer at all.

The process described for the +2EV layer is repeated with the -2EV layer. I only want to retain bits of the water from this layer, meaning that most of this layer will be deleted through the layer mask being brushed 'black'

You can repeat the procedure with new layers for other exposures that you took if there is a particular aspect of each exposure whose appearance you want in the final image. An example of this is for waterfall images. Because of the way I shoot these images, the best water motion is often contained in the 0EV image which I will add as an additional layer, removing everything else (using a layer mask) but the water I wish to retain in the final image.

The end result should be that you have an image with good dynamic range where shadows are visible and highlights are well controlled. By blending in the original images , the resultant image looks more real than the original image that Photomatix produced for you.

The final result of melding the three layers together. Now all that's left is to do is"flatten" the image

Once you are happy with the exposures and general appearance, ‘flatten the image’ and then whatever workflow for all your non HDR images comes into play….but that’s a post for another day!

So , just to show you the evolution again , We went from

The three native images:

To a Photomatix generated image : (weird lighting)

To a more natural looking image by blending in the original images: (soft looking image)

And finally, some more adjustment to colours, levels, sharpening layers and borders for the final result:

Hooray all done!!!

Well, that was probably a bit of a lengthy read but I hope it would at least have given you some ideas as to how to to about producing these kinds of images after taking the shot!(bigger version at http://www.flickr.com/photos/dmtoh/4142172567/)



Posted on November 29, 2009, in China, How we..., Photography and tagged . Bookmark the permalink. 6 Comments.

  1. Great tute – thank you 🙂

  2. Hello. finejob. I did not expect this website on a Mon.

  3. Just a fantastic tutorial. When I was about to step away from HDR all together, I find this. Now I am tempted to play again! Thanks!

  4. I have been exploring for a little for any high quality articles or weblog posts on this kind of area . Exploring in Yahoo I eventually stumbled upon this site. Studying this info So i am glad to show that I have an incredibly good uncanny feeling I found out exactly what I needed. I most certainly will make sure to don?t disregard this website and provides it a look on a continuing basis.

  5. Thanks for sharing this, it was very helpful.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: