Capture More Light: How to Capture High Dynamic Range (HDR) Images with your DSLR

March 28, 2011

… there’s a little more to getting good HDR results than just blending two images together. Simply combining two or more images taken at different exposure levels often isn’t enough to produce an excellent, finished HDR image. 

This is the first article of a 3 part series:

  1. HDR and how to create HDR images
  2. High speed bracketing method, created by the author (click to read)
  3. Getting the most out of your HDR images (click to read)


What is this talk about HDR all about?

High Dynamic Range (HDR) photography is something you hear digital photography enthusiasts speaking about a lot these days.  Basically, it means combining two or more images — of the exact same scene — taken at different exposure levels into one finished image.  One dark exposure can capture detail in bright areas of the scene, while a second, lighter exposure can give details in mid-tones and shadows.  Combined, the two can give a finished image with far more overall tonal detail than is sometimes possible to capture with even the finest digital SLRs on the market today.  This is especially true in very high-contrast scenes, such as scenics in bright sunlight, and indoor shots taken in the daytime with bright windows in the scene.

As old as photography is the search for the ideal light. Part of this search for light is related to the fact that some scenes just look better in certain (often warmer) light. But even more it is related to the fact that film and now modern digital DSLRs can only capture a certain dynamic range.

The Dynamic Range (DR) of a DSLR or film is measured by the ratio of the brightest detail in the highlights and the darkest detail in the shadows. The detail in the highlights is limited by pure highlight clipping and the detail in the shadows is limited in modern DSLRs by the noise level (or film grain).

It is common to measure the DR of cameras in the range of Exposure Values (E.V.) spanned, E.V. being often expressed as f-stops (even though photographers usually vary the exposure time rather than the aperture. Some data can illustrate this:

Examples of Dynamic Range

Black & White Negative Film

10-11 f-stops,  or a DR ratio of about 1,000:1 – 2,000:1

Slide Film

6-7 f-stops, or a ratio of about 100:1

DSLRs (in 2008)

8-10 f-stops, or a ratio of about 250:1  to 1000:1

Daylight Scene (with full sun)

12-15 f-stops, or a ratio of 5,000:1 to 50,000:1 (depending on preferred amount of shadow detail)  

Room Interior, with outside view from window into full daylight

At least 17 f-stops, or a ratio greater than 100,000:1


Here is the situation: The scene can have often a larger DR than the camera can capture, and the camera can capture more DR than we often can print.

Exposure Merge

Of course all three options are valid. This article concentrates on the last option. How can we overcome the limitations of our cameras. Because the problem is as old as photography we also had solutions for as long. We capture two or more separate photographs (from underexposed to overexposed) of the same scene and combine the images into a new photo. The classic solution is called Exposure Blending. Master photographers did this manually in the darkroom with enlargers, and today the same is done in the digital world in Photoshop™. Seamless manual blending is hard work because of the issue of:

  • Aligning the source images
  • Masking the images to get seamless blended results

Most digital photography enthusiasts’ first efforts at combining separate images into one finished HDR image is done in an image-editing software program such as Adobe Photoshop™.  Using features such as layers and various masking techniques, it’s possible to get excellent results, if the user’s technique from the camera to the computer has been sound.

However, today there are quite a few specialized software applications (one that I’ve had excellent results with is Photomatix Pro) that can are designed primarily for the single task of merging two or more exposures into a single image, which is LDR (Low Dynamic Range), but containing the details in highlights and shadows of the exposures from the sources images. There are different merging algorithms with different strengths and weaknesses.

HDRI (High Dynamic Range Imaging)

HDR Imaging is actually something new. We don’t want to confuse you with too much detail, since it is not needed for practical HDR work. HDR allows you to store image data that capture a nearly infinite tonal range (from the deepest dark areas to open sunlight). The HDR images store image data as 32 bit floating point values in order to encode real values of light.

Instead of merging different exposures into a single LDR (normal JPEG or 8/16 bit TIFF) image, you can create a HDR image that can represent all the tonal values in even the most contrasty scenes.  Applications like Photoshop™ and Photomatix Pro can create such HDR files from a series of different exposures.

Tone Mapping Methods (Operators)

However, there’s a little more to getting good HDR results than just blending two images together. Simply combining two or more images taken at different exposure levels often isn’t enough to produce an excellent, finished HDR image.  A software process called “Tone Mapping” further alters the contrast range, giving a vibrant and deep range of tones that’s suitable for our computer monitors and printers. High-end image-editing software such as Photoshop™, and many of the dedicated HDR software applications such as Photomatix Pro also handle this step. Both applications implement very different Tone Mapping methods (operators). Right now the Photomatix Pro Details Enhancer method seems to be a  popular one. Over the next years other methods will likely appear.

Capturing photos for HDR

Since my primary software for combining images into HDR files is Photomatix Pro 3.0, we’ll briefly demonstrate the process using this software (this program sells for about $100;  free demos are available at www.hdrsoft.com). This is by no means a tutorial on Photomatix but the process in Photomatix is  easy to learn.  Again, there are also other HDR-dedicated programs you may want to look into;  furthermore many users start by combining two or more images with image-editing software such as Photoshop, and this is perfectly effective as well.   For this article it is our intention to show the principle three-step HDR workflow:

  1. Capturing bracketed exposures
  2. Generating the HDR image
  3. Tone Mapping of the HDR image

Different exposures of the same scene are best created at 2 EV spacing (e.g. -2 EV, 0, +2 EV). All current Canon DSLRs support an AEB (Auto Exposure Bracketing) modus where the camera handles the proper bracketing and also allow at least 2 EV steps.

For normal outdoor scenes, three shots at -2 EV, 0, +2 EV are often sufficient. The camera should be used in Aperture priority mode. This means the camera handles automatically the different exposures by changing the shutter speed.

Movement is the enemy of HDR. Why? Think of combining images where certain parts are at different locations in the different exposures. This is called “Ghosting” because the combined image would show these elements like ghosts. This is why most HDR photographers use a sturdy tripod and all techniques available to minimize camera shake (e.g. use the “Mirror Up” feature in your camera). Keep this in mind, because in our Part 2 we will show a technique that works also freehand — and changes the whole character of shooting HDR images. Slight movements of the camera can be compensated in the HDR software (even photo series from a tripod are often not perfectly aligned).

We demonstrate the HDR workflow on one of our Alcatraz portfolio photos (taken with Canon EOS 1Ds Mark II + Canon EF 24-70mm f/2.8 L USM lens on a sturdy tripod, mirror was up and a cable release used).

Here are the three different exposures (original Canon RAW files):

Processing step #1: Generate HDR image

In Photoshop CS3, the equivalent function is called “Merge to HDR”  (File > Automate > Merge to HDR).  As mentioned previously, there are alternate methods in image-editing software that are more user-intensive, involving masking and combining images manually as multiple layers into one finished image.

Here’s a quick view of how it’s done in the specialized Photomatix software, which is a program designed specifically for this purpose.  (Other similar software applications will usually offer similar features and steps.) We select our three exposures in Photomatix and have to select some crucial options:

Most of these settings are default values. We nearly always enable the “Align source images” option. There are two different algorithms available. We start most often with the method “By matching features”. Only in some rare cases if this fails to work we also try the other option. Once we click ok Photomatix does its job and creates the initial HDR image.

If you see a HDR image for the first time, its excessive contrast comes as a shock:

Why is this? Remember that HDR images can represent a dynamic range that no normal monitor can display. This is what you experience here and that is also why the next Tone Mapping step is essential. If you plan to try different Tone Mapping settings or algorithms this would be the time to save the initial HDR image. The images are saved in a file format that stores the extremely broad 32 bit image data. Popular file formats are: 

•    Radiance: uses a .hdr suffix (our choice)
•    OpenEXR: uses a .exr suffix (good alternative)

Processing step #2: Tone Mapping

Although Photomatix offers two different Tone Mapping algorithms we nearly always use the Details Enhancer. The Details Enhancer performs a tone mapping based on local contrast optimization.

There are quite a few options (sliders) to tweak to get the optimal result. Because this is not a Photomatix tutorial we  won’t cover this in detail here. In the end it is not that complicated and you will use similar settings over and over again. If you hit Process, Photomatix generates a tonemapped 16-bit image.  Now, the contrast range will seem much more appropriate for the scene and its lighting.

In our work the tonemapped image (saved as 16-bit TIFF) is never the final result. We always use Photoshop™ for final retouching and sharpening.  Even though most final, tonemapped HDR images are color images, in this particular case, we opted for a B&W version:

Summary of Part 1

Capturing multiple images with your camera and combining them into finished tonemapped HDR images is quite simple today. Getting the right artistic results will as always take some time practice and experimentation. In Part 2 we cover our “High Speed Bracketing” technique which builds on the technique you learned in Part 1.

We conducted some workshops on HDR called “Capture more Light”. All our students found that working with HDR is fun because it adds a new tool to your creative process.

Final note of caution

We mentioned it in this article: Movement is the enemy of HDR. This means that you have to watch for everything that moves. Here is a list of common ghosting* candidates:

  • People
  • Cars, bicycles, and other vehicles
  • Larger animals (dogs, cats,...)
  • Small animals (birds, butterflies,..)
  • Clouds
  • Water
  • Flags
  • Trees and plants in wind (means nearly always problems)


* In Part 3 we will talk a bit more about ghosting. 

The CDLC contributors are compensated spokespersons and actual users of the Canon products that they promote.

All images are copyright Uwe Steinmuller

http://learn.usa.canon.com

© 2014 Canon U.S.A., Inc. All Rights Reserved. Reproduction in whole or in part without permission is prohibited.