The Fujifilm X20 and Dynamic Range Expansion

There's a well-known digital technique called "Expose to the Right" for working with scenes that have high dynamic range.  The technique is somewhat controversial (read the link if you want to know more about that).  I don't recommend it unless you're dealing with a high dynamic range scene and you know upfront that you're going to post-process the photo.  But when it's done correctly on a scene that actually needs it, it will yield an image that's optimized for compressing the dynamic range in post.

The technique works on the principle that a RAW file can retain a remarkable amount detail in the shadow areas, but a blown highlight is gone for good.  Exposing to the right means you push the exposure as high as possible but stop short of "blowing out" any area. This means that the histogram will show pixels bunched up on the right side of the graph, but with no pixels actually pegging the far right end.  Then, in post-processing the photo, you lighten up the mid-tones and shadows to reveal the detail there.  In that way, you can get good detail in the shadow as well as the highlight areas.   

The Fujifilm X20’s "dynamic range expansion" settings basically automate this procedure.


Here's dpreview’s description of dynamic range expansion in the X-Pro1, which is the same as is used in the X20 (the emphasis in bold is mine):

The X-Pro1 has two dynamic range expansion settings to bolster its highlight range: DR200 (200%) adds an extra stop of information in the highlights, and DR400 (400%) adds two stops. Technically, the camera achieves this by applying less amplification to the sensor's output than usual prior to AD conversion to avoid clipping highlight data, then pulling-up the midtones to the correct brightness in JPEG processing...

An alternative way of thinking about this is that DR200 is like underexposing a stop to retain highlights then adjusting the brightness afterwards, and DR400 is like underexposing by two stops and adjusting further.
http://www.dpreview.com/reviews/fujifilm-x-pro1/20

So, when the X20 underexposes to retain highlights, it does so by lowering the amplification of the sensor’s output.  It’s not clear from the dpreview’s description, but that’s the same thing as lowering the ISO.  In case you don't believe me on that, here's an excerpt from Wikipedia’s description of how digital cameras implement ISO settings (a.k.a. exposure index or EI rating).  Once again, the bold emphasis is mine:

For digital photo cameras ("digital still cameras"), an exposure index (EI) rating—commonly called ISO setting—is specified by the manufacturer such that the sRGB image files produced by the camera will have a lightness similar to what would be obtained with film of the same EI rating at the same exposure. The usual design is that the camera's parameters for interpreting the sensor data values into sRGB values are fixed, and a number of different EI choices are accommodated by varying the sensor's signal gain in the analog realm, prior to conversion to digital.
http://en.wikipedia.org/wiki/ISO_sensitivity#Digital

“Applying less amplification” is the same as reducing the gain.  So, when the X20 underexposes to retain highlights, it does so by effectively lowering the ISO.  So even though you had to set the ISO to 200 to use DR200%, or ISO 400 to use DR400%, the actual ISO employed is 100.  (By the way, the EXIF data will show the higher ISO value that you dialed in, but reducing the sensor gain is effectively lowering the ISO no matter what the EXIF data says.)  I've seen a lot of posts on the Internet saying that the reason the Fujifilm dynamic range expansion process results in more noise in the shadows is because the ISO was raised. But based on the fact that it is underexposing by effectively lowing the ISO, I don't believe that is true.  I posit that it’s from the brightness correction to shadows and mid-tones that occurs during JPEG conversion.  The same thing happens when you brighten up shadow areas using Photoshop or Lightroom as well - you amplify any noise in those dark areas.

If you're familiar with the X20, you may be wondering why the X20 forces you to use a higher ISO with the DR settings.  Good question.  I'm pretty sure it is simply so that it has room to lower it for the dynamic range expansion!  It first uses the ISO you dialed in to calculate exposure (and set shutter speed or aperture if you’re not in M mode), then it reduces the ISO in order to underexpose it for DR processing.  Seems a bit convoluted, but it makes sense if you think about it.  Unless I need to lock down my ISO speed for some reason (like I'm shooting HDR brackets or I'm using a tripod and prefer long shutter speeds over high ISO, for instance) I usually set ISO to Auto, and in that case, the X20 will automatically set the ISO to the appropriate value for the DR settings.

In practice, I had been just leaving the camera set at DR200%.  It seemed like a nice way to protect against blowing out highlights without having to think about it.  But I've changed my strategy to leaving the camera at DR100% (the lowest setting) and only using higher DR settings when the dynamic range of the scene actually warrants it.  Why?  Well, the DR settings are kind of crude in how they work.  If you were exposing to the right manually, you'd underexpose just enough to avoid clipping and no more.  But the DR setting underexposes by 1 or 2 stops regardless.  So in most instances it's probably going to overdo it a bit and generate more noise than is really necessary.  The small sensor in the X20 is already susceptible to noise in low light, so my current thinking is I'd rather not make it worse unless I really need to.  On the other hand, the DR settings are a heck of a lot easier and faster than doing it manually, so I can definitely see a lot of usefulness with them if you need to deal with high dynamic range but can't spend a lot of effort and time to do it.