In case you are curious what I’m dealing with in this article, have a look:

HDR of Orthodox Cathedral, Sibiu, Romania

The cathedral is located here.

If you find it interesting, read on.

Introduction

Just in case you’re unfamiliar with the therm, HDRI stands for High Dynamic Range Imaging. As I’ve discussed before, a regular photo can’t capture the full range of lights and shadows that the human eye can see. Photographers have long learned to avoid high-contrast situations or to use them to create artistic effects. It all changed when Paul Devebec presented at SIGGRAPH 1997 a method for combining several images with different exposures into one single image with a much higher range of luminosity. To put things in perspective, a camera sensor can cope with contrast ranges on the order of 4000:1, while a scene of a room with an outside view in full sun (something our eyes see every day) has a dynamic range of 100,000:1.

Initially HDR was used in 3D graphics as environment maps to create realistic scenes. Because a HDR image contains 32bits per channel (in floating point, e.g. can take almost infinite values), it cannot be displayed directly on the screen or printed on paper, so a second operation is needed, called Tone Mapping. Tone Mapping consists in “compressing” the high dynamic range back into something we can see on screen. This is something easier said than done; a simple conversion results in a lifeless, murky image so there’s a whole research field in creating algorithms that present images that are realistic and pleasing for the eye.

Shooting the scene

Equipment used:

• Sony α700 DSLR;
• Sigma 10-20mm lens;
• tripod (Canon, in case you’re interested).

First, a straight JPEG of the cathedral interior is hopelessly bad. The stained glasses are blown out while many areas are pure black. Colors are muted and overall it’s not an incredibly attractive picture. Shooting with Dynamic Range Optimizer marginally improves the shadow detail, but not by much.

JPEG image lacks the needed dynamic range

So, I put the camera on a tripod and shot eight pictures, from under- to over-exposed, one f-stop apart each. I started with just enough light to see the windows and ended with an exposure long enough to see clearly in shadows. Here they are:

8 images with different exposures make the final HDR

Picturenaut

Picturenaut is a nice little FREE tool from a guy named Christian Bloch. The interface is pretty barren, but it has to be the fastest HDR tool I’ve tried. Did I mention it’s free? (some tools can cost $700)

Selecting images for HDR

Tone Mapping dialog

First step is selecting all images that will make the final exposure. Picturenaut has the option to automatically align the images, which is extremely important since even with the camera on tripod the image can be slightly different from shot to shot.

Aside from being very accurate, Picturenaut is really, really fast. It generates the HDR in no time. The HDR image displayed on screen usually looks rather flat, but that’s expected. You can save it as a 32bit TIFF if you want.

Second step is the “fun” part – tone mapping. Picturenaut offers two algorithms (Adaptive Logarithmic and Photoreceptor physiology). Which one is better depends on the actual scene.

The tone mapping is also very fast, almost real-time.

Photoshop

Image selection for HDR

Tone Mapping in Ps

You access the HDR option from File –> Automate –> Merge to HDR (rather unintuitive, but that’s me). The process is very slow, “Aligning layers based on content” message seems to take forever.

For Tone Mapping (achieved via Image –> Mode –> 8 bits/channel), Photoshop offers four options: “Exposure and Gamma“, “Highlight Compression“, “Equalize Histogram” and “Local Adaptation“. Again, depending on the scene, one mode may be more useful that the other, but I find Local Adaptation to be good, with Highlight Compression and Equalize Histogram essentially useless.

Image Quality: Photoshop vs. Picturenaut

With both programs, after tone mapping, I spent some time tweaking the image, playing with Curves to increase the contrast and Saturation to bring out the true colors I saw in the cathedral.

Here’s the side-by-side end result:

HDR result - Photoshop vs. Picturenaut

Apart from some color differences, the two images are remarkably similar. I did manage to preserve highlight detail better in Photoshop and its local contrast made everything “pop” more, but overall they are on par, so if you can’t afford Photoshop or some “pro” tool, Picturenaut will deliver good results (but you still need to tweak it, it’s not a “push the button and you’re done” kind of operation).