Tuesday, July 24, 2012

HDR Photography

High Dynamic Range (HDR) Photography has long been popular, especially with the realization that early digital cameras had a rather limited dynamic range, and various means had to be created to correct this problem. The JPEG Digital File Standard remains only at 8 bits, which only allows 256 distinct shades of light to be processed (thinking strictly in terms of black and white). More recent cameras do their processing with 14 bits, which increases that number dramatically to 16384 values, which provides much smoother transitions from dark to light, and when combined with shooting mode options like Canon's “Auto Lighting Optimizer”, this extended range can be used by the camera to increase shadow detail. However, conversion down to JPEG will ultimately reduce the range back down to 256 values. If you convert images to the Tagged Image File Format (TIFF), which can accommodate 16 bits instead of JPEG, the range of values your camera used to process an image will be retained.

This has little to do with the technique that evolved from this need to improve digital pictures however – the above merely describes where the HDR technique came from, as in “necessity is the mother of invention”.

Instead of explaining the necessity in terms of numbers, I'll explain it in terms of what the eye can see (which I should have done in the first place, right?) In a lot of photographs, (even when done with film), there are two really big exposure problems – details that are lost in the shadows, and details that are lost in the highlights (“blown highlights”). Processing with a greater than 8 bit depth can help diminish these problems, but the HDR technique takes it even further. With HDR, you take three or more of exactly the same picture, preferably with the camera on a tripod and absolutely nothing moving in the scene. These pictures must be all exposed differently, with the lowest exposure showing all the shadow detail, and the highest exposure showing all the highlight detail. These photos are then “stacked” using Photoshop or equivalent software, and then blended and combined into one image that shows all the detail – from deep shadow through to the brightest light. What you get is a strange but often wonderful looking piece like my shot of St. Alban's Church above.

My opinion of HDR is that it has become a branch of the art of Photography. I think that HDR Photographs look a lot like paintings, because a painter will often brush in all of the detail he is seeing, as his eye adapts from dark to light within the scene he is painting. But here's the thing – our eyes do not see all this detail at one time. Our pupils must adjust to the light as we alternately peer between deep shade and bright light. Yes, our eyes discern all the detail at both extremes, but like a camera, our eyes don't see it all at one time. This is why an HDR photograph, or a lot of paintings (depending on what the artist intended), tend to look “hyper-real”. I should say that the best HDR photos look that way, but most of what I've seen simply look “flat and over-bright”. They lack in depth, because all of the contrast that our eyes can take in within one spit second has been eliminated from the photograph.

Let's consider it in yet another way. High Dynamic Range is essential in Cinematography, because a movie camera is constantly moving from dark to light in much the same way our eyes do. A really crappy movie camera might not respond to these extreme lighting changes, and create movies that look dull because a lot of deep shadow and bright light detail is missing. A good movie camera, on the other hand, will respond correctly to the changes as it moves through a scene. But, like the human eye, a movie camera never has to see deep shadow and bright light detail all at the same time, and yet, this is exactly what the HDR photo technique is seeking to do.

Finally, I will mention a couple of other, and perhaps more suitable ways to make the dynamic range of a photograph stand out.

  • Use a Graduated Neutral Density Filter on your camera lens, if you're shooting a traditional “bright at the top, dark at the bottom” landscape. These screw-on filters darken the top half while leaving the bottom half of the picture in full light, thus reducing blown out highlights, allowing more highlight detail to come through, especially in the clouds.
  • Convert your pictures from RAW to TIFF instead of JPEG. This will retain all of your camera's efforts to keep graduated light detail intact. The downside is that TIFF files are many times larger than JPEG's. I would recommend this only if you intend to print the image on a TIFF capable printer. (Note – this will not create the “HDR-look” - it will still look real, but slightly better)
  • An HDR look can be achieved with just one picture – you don't need to combine three or more. Because you are going to collapse the whole output of the HDR processing down to 8-bit JPEG anyway, you can still get the same look of an HDR by using your RAW conversion software to 1) brighten the picture, 2) apply “dynamic range compression” 3) apply “local contrast” to the mid-tones, and 4) de-noise the shadows. This is what I did to the church photo above, using Photivo as my RAW sotware. The picture started out looking like this:

Some people call this “Fake HDR”, which I find kind of funny, because of the fact that the human eye does not see all levels of detail in one instant, then HDR is all fake anyway, no matter how it's done. With HDR, you gain lots of lost detail, but, mainly because of the final compression back to 8-bit, you lose a lot in the way of depth suggesting contrast (or Luminance). I think to make a good HDR project requires a light-handed touch.

1 comment:

Reader's comments are welcome, and are subject to moderation by the author.