What Is Dynamic Range?

Poking about the web yesterday for info on the Fuji F200 EXR, I came across the above illustration on Fuji's website, showing example pictures from the F200 without (left) and with (right) the camera's "D-Range Priority" feature switched on. (Click on the image to see it larger.)

It struck me that this is also a very nice illustration of what dynamic range is, for those who might not be quite clear on the subject. (I always tend to like visual illustrations of visual phenomena best. One thing I've always meant to do is to put together a compendium of image qualities and properties in visual form; yet another thing to do that will probably never get done.)

The new EXR sensor is said to differ from ordinary image recovery and DR-maximizing modes in that it (allegedly) captures an image simultaneously in both its different capture modes and combines them—something more similar to two-shot HDR schemes than what's usually available with in-camera features.

Mike

Send this post to a friend

Comments

[michael]

What anyone achieves with the Shadow/Highlight command in Photoshop.

[Seinberg]

That's actually really impressive, if it's legit. The pillars look like they had their contrast increased in post-processing, don't they? I want to give them the benefit of the doubt, because it's a really novel idea that will be standard on most cameras eventually, but it just seems like it's a little *too* much of a difference.

[John Sartin]

My initial reaction to this comparison is that I actually like the shot with the sky blown out more then the more "dynamic shot". That said, I would always prefer to have the range available if I needed it.

[Jasiek]

Hi

I recently upgraded from a dying D70 to brand new S5 PRO.

And what the image show is true and achievable. You can safely expose for shadows, and highlights will be there anyway. Unles you happen to shoot straight into the sun, that is...

Sure, this camera has some difficulties. Basically, most of Fuji-developed areas are problematic to use (awkward menu, slow processor, etc...). But with this great sensor and Nikon-made body (body controls thankfully unchanged from D200) it is a fantastic camera to me.

I wonder if this camera may solve Mike's dissatisfaction with highlights in digital B&W.

Had I lived closer, I would seriously consider lending it to Mike just to see his opinion.

[Jim Allen]

Hi Mike,
What is F-stop? Here's a visual image to illustrate that.(at least the way it's adjusted.)
http://www.flickr.com/photos/26317521@N00/1410041461/sizes/o/
For your collection of images of phenomena article...
Jim A.

[Mike Johnston]

"What anyone achieves with the Shadow/Highlight command in Photoshop."

Michael,
Well, not according to Fuji, no. Although naturally we'll have to wait until the camera is released and reviewed before we know for sure.

The first picture would be the equivalent of one in which the histogram was cut off on both the left and the right. The second picture would be one where the entire histogram fits within its box. You could make the left-hand picture from the right-hand file, but not vice versa.

Again, whether this is really true in practice awaits actual testing.

Mike

[improbable]

"What anyone achieves with the Shadow/Highlight command in Photoshop"

No, the idea is to capture more dynamic range, not just to alter how what was captured is displayed.

But this does raise an interesting question: after capturing this HDR image, some processing is needed to map back into the normal range, since the camera wants to produce an email-ready jpeg. This second step is what shadow/highlight is about.

In the Czech samples posted above, I can't see any of the artefacts one sometimes sees from tone-mapping, halos around trees etc, which is a good sign. Although looking at the full-size pictures, the detail seems much more smeared (plasticky) on the HRD versions of the shots. Perhaps this is because the non-HDR ones are like downsampled 12MP shots from the point of view of noise reduction & demosaicing?

[Lambert]

This particular example looks a bit unnatural. I'm not sure whether it extends the dynamic range, or just "flattens the contrast", for lack of a better term. At any rate, I'll be taking a closer look at this camera, mainly for the manual mode and IS, as my trusty old pocket camera has given me "System error" a couple of times already.

[mikeinmagog]

The right one looks thin at first until you allow yourself a look at the proper blue sky in the upper left. I might've tweaked it a smidgen 'blacker' but that's just me.

And, michael, achieving this straight out of the camera is much preferable to coaxing with Photoshop, where you would risk stressing out those poor pixels.

Onward, digital photography!

[Geoff Wittig]

I've been doing a lot of HDR photos over the last year, as insane contrast ranges are the rule with backlit sunrise landscapes. The end result is often wonderful, but the process is mighty tedious, what with bracketed captures and serious computer time creating a realistic rendering of the blended files. I have to think it would be a simple matter of firmware programming to get any decent D-SLR to do most of the work for me. The first camera maker to include a good, workable HDR capture mode coupled with automated rendering in their Raw converter software would surely capture my attention.

[Paul Amyes]

Geoff Wittig wrote

"The first camera maker to include a good, workable HDR capture mode coupled with automated rendering in their Raw converter software would surely capture my attention."

Please let this happen. I've gone back to shooting medium format colour neg for my personal landscape photography, but all that scanning is a real pain. Bring on a digital solution.

[mike]

OK, so I'll be the one to ask the question: how would the exposures differ on those two examples? What is the photographer on the right doing that the one on the left isn't in order to achieve the result?

[Jeff Hartge]

I have a Fuji S3 Pro. One of the reasons I purchased this camera was to have more headroom for creating a film-like shoulder on highlights.

At the time I purchased this camera, most of the "DX" cameras available didn't have sufficient headroom. Now, all of the newer cameras have more than enough headroom to get nice roll-offs (8-9 stops) in almost any type of image.

At this point, most compact cameras seem to have around 5 stops of DR. That doesn't give much wiggle room to avoid clipped highlights.

The F200 EXR will be useful if it can come close to the same dynamic range as the current crop of DSLRs. To have 7 stops plus a little more for roll-off would really cover most "standard" photographic situations admirably.

[Mike Johnston]

"how would the exposures differ on those two examples? What is the photographer on the right doing that the one on the left isn't in order to achieve the result?"

Mike,
The best answer so far can be found in the .PDF download in the "Notes and Rumors" post.

Mike J.

[Polar Plunge Guy]

"One thing I've always meant to do is to put together a compendium of image qualities and properties in visual form"

Maybe you should think about finding some online group of people that would help you... you could basically be a facilitator and editor for contributions illustrating particular properties. But where would you find such people?

[Dr. Nick]

Mitten in ein DSLR, pleasen?

[Wolfgang Kuechle]

Well, for "the old folks" who still know what a darkroom is, it may also be explained by the analogy of papers of varying gradation. Essentially the F200 offers two sensor modes of different gradation "normal" and "soft".

[António Pires]

If we are opening up the shadows and detailing the highlights it means that both will take a larger share of the 256 levels of the response curve. It means that the mid-tones will have less levels to expand to, i.e. the mid-tones will be less contrasted, giving the impression of a flattened contrast as Lambert noticed.
The alternative is a sort of jig-saw curve, like the HDR technique, where shadows, highlights and mid-tones are all converted to mid-tones.

[Renato Aranghelovici]

Jeff, F200EXR is speced by Fuji as having 11, yes eleven f-stops in that DR special mode!

[improbable]

António Pires - Yes, if you squeeze an image with greater-than-normal dynamic range into a standard jpeg, it'll look very flat. But HDR tone-mapping doesn't use a "jigsaw curve", rather it treats different parts of the image differently, trying to maintain local contrast (within a dark area) while reducing global contrast (between the sky and the shade).

Most of the HDR pictures found on say flickr have this process turned up to 11, to make freakish pictures without any contrast at all. But more subtle versions of it are, I think, a better representation of how our eyes see the world than is straight "linear" photography. They will, I think, become more common as sensors get better.

Incidentally, this process is something like audio compression (in which the volume is adjusted from one second to the next, based on loudness) just in two dimensions. Contrast masking (as of cibachrome) is also much the same thing. And I think that digital minilabs have done this automatically for years.

[Rana]

I'm intrigued, though I'm not giving up my Pentax any time soon. Shooting with the Pentax (an *istDS) has been much like shooting slide film for me - I have to be careful to not blow the highlights when I'm exposing for the darker areas of the subject.

Photoshop does allow some manipulation of the darker areas, but when all you've got are white pixels to work with, there's little you can do. (The same's true of solid black areas too, of course, but the highlights seem to be more prone to that kind of reduction than the shadowed areas. And, yeah, one can shoot two exposures of the same thing and combine them, but that doesn't work very well with active subjects.)

It would be interesting to work with a camera that not only had an extended dynamic range, but which could toggle between an "extended" and a "regular" mode, as this one seems to do.

[David Dyer-Bennet]

I'm not sure which one I prefer -- which makes them a perfect example, really.

The second one contains both shadow and highlight detail missing from the first, just to state the obvious; that's why it's said to capture "more dynamic range", and that's why you can't get the same effect just by altering the exposure.

The visual encyclopedia of photographic concepts seems like an *excellent* concept; I would attempt to produce some of the necessary illustrations, if somebody was driving.

[Jim]

In other words, there ain't no free lunch.

[Phat Photographer]

You have to hand it to Fuji for at least pushing the envelope on sensors and attempting to deliver real value. Having the option of more dynamic range to me is far more interesting than just cramming more megapixels into the sensor. If I recall correctly, there's also an option to shoot at a lower resolution with less noise by combining the light gathered from adjacent photo diodes as opposed to taking two separate exposures and combining them.

[Rana]

Oh, and add me to the list of folks who would be interested in such a visual encyclopedia.

[Jeff]

I believe Nikon calls it D-Lighting

[Mike Johnston]

"I believe Nikon calls it D-Lighting"

Similar effects, entirely different technology.

Mike

[David Bostedo]

"Similar effects, entirely different technology."

I would argue that it's a different effect and technology. D-Lighting will give you head room and keep from blowing the highlights, but only by underexposing and pulling up the shadows after the fact. You could do the same thing by underexposing youself, and using Photoshop on the shadows.

What people may not be getting is that things like the Shadow/Highlight tool and D-Lighting don't actually give you the ability to capture any more dynamic range (DR). DR is determined by the properties of the sensor - not the exposure, not the post-processing, etc. Other techniques may emulate it by boosting what you want at the expense of something else, but there's no actual added DR.

I also wonder how much of people preferring left image is because that's what they're used to seeing in photos. The right image (I would guess) is much closer you what your eye would see - not that all photos should try to match your eye or anything.

[Kevin]

"I also wonder how much of people preferring left image is because that's what they're used to seeing in photos. The right image (I would guess) is much closer you what your eye would see"

In the real world your eyes jump around scanning different parts of whatever you're looking at and your brain merges them, with the emphasis very much on the part of the scene that you're most interested in at any given time. Your eyes and your brain are continually adjusting to maximise your perception of the part of the scene you are focused on at any given moment; you don't much notice the peripheral areas unless something happens to draw your attention to them.

When you look at an an HDR image you're forced to take in the whole scene in one hit, which isn't a natural way of looking. Your visual system would probably process a mural-sized HDR image similarly to the way it processes the real world, but when the HDR image is a typical-sized print, magazine illustration or computer monitor display it just looks odd.

[Javier Guillén]

Not really related to the subject of this debate, but... The picture is of the "Plaza de España", in Seville, Spain. I walk through that precise spot everyday on my way to work. And now everytime I do so think about T.O.P. (!!)

Anyway, cheers from Seville, and thanks for the good work.

Signed: A long-time reader.