DSLRS and dynamic range and bit-depth in photoshop

[jenne1978]

Hi there and thanks for taking your time reading this post.

I have two questions that seem quite hard to get an answer to, I hope due to errors on my part.

1: is there some kind of rating website or anything that gives me an insight in what the actual dynamic range is of DSLRS? It seems very important to me to know if there is a dynamic range difference between, for instance, a Canon 30D and a canon 5D MkII. Note: I mean the actual dynamic range of the raw file, not the bit depth of the A/D converter. Those are always listed with all the other specs of the camera.

2. My second question is somewhat related to the first. Why is it that in 16-bit mode Photoshop still displays 8-bit figures in the info window? One would expect that after conversion the whites would be displayed as 65535 for all the channels, but instead all the channels still show 255. I suspect this is because of limitations of the display and the human eye, but I really am in the dark here (no pun intended).

Regards,
Jenne Hoekstra

[MrB]

1: is there some kind of rating website or anything that gives me an insight in what the actual dynamic range is of DSLRS? It seems very important to me to know if there is a dynamic range difference between, for instance, a Canon 30D and a canon 5D MkII. Note: I mean the actual dynamic range of the raw file, not the bit depth of the A/D converter.

Snapsort list a value for dynamic range but I don't know if this is the one that you want -

http://snapsort.com/compare/Canon_EO...EOS_5D_Mark_II

Philip

[tclune]

1: is there some kind of rating website or anything that gives me an insight in what the actual dynamic range is of DSLRS? It seems very important to me to know if there is a dynamic range difference between, for instance, a Canon 30D and a canon 5D MkII. Note: I mean the actual dynamic range of the raw file, not the bit depth of the A/D converter. Those are always listed with all the other specs of the camera.

The standard place to go for detailed evaluation of camera dynamic range (among many other quantitative measurements) is the DXOMark website. Along the top of this link: http://www.dxomark.com//index.php you will see a menu item, "Cameras." Put your mouse on it and you will get a drop-down menu that includes the "Camera sensor" database; select the manufacturer you are interested in, and you will see a long list of camera models that they have evaluated. Select the camera you are interested in, and then you will see a tab labelled "Measurements." The graph you want is labelled "Dynamic Range." You will see a graph of EV vs ISO. "EV" in this case actually could be replaced by the word "bits." So, if the camera sensor is 14 EV at ISO 100, it has 14 bits of dynamic range at that ISO (assuming, of course, that you have exposed the image to use the full dynamic range). The DXOMark website is a veritable wealth of authoritative technical information on cameras and lenses, but it is about as unfriendly a site to navigate as you'll ever see (in case you hadn't realized that already).

[Dave Humphries]

Snapsort list a value for dynamic range but I don't know if this is the one that you want -

http://snapsort.com/compare/Canon_EO...EOS_5D_Mark_II

Sorry Philip,

My personal opinion of snapsort is "avoid it".
It is riddled with inaccuracies and mis-leading/irrelevant statistics like no other site I have ever seen.

Case in point from this very page, in the Table of advantages of Canon 5D Mark II;
"More dynamic range 11.9 EV vs 10.8 EV 1.9 f-stops more dynamic range" (my bold)
... errm, I make the difference between 11.9 and 10.8 to be 1.1 EV, anyone disagree with me?

Every single time anyone has linked there I have found silly conclusions and silly errors, makes Ken look good

It is so bad I just cannot condone it, sorry.

... and Yes, I have just told them about it (so I hope I'm right)

[Colin Southern]

Hi Jenne,

Tom's dxomark.com is the "A" answer, but in reality, dynamic range usually doesn't make any difference to everyday photos because although the camera captures around 12 stops (which is typical at base ISO), monitors only display around 6 (and prints around 4) - so most of it is discarded (even after a stop or two DR compression to reveal shadow detail etc).

About the only time I push my camera to the limit DR wise is shooting into the sun and trying to preserve shadow detail ... and even then it's accomplished more cleanly with a GND filter (because the 2 or 3 stops closest to the noise floor are generally very noisy).

Hope this helps.

PS: Photoshop only uses 15 bit processing (not 16 bit) - not sure why they don't give 16 bit numbers though - but keep in mind that the human eye can only descern about 200 level changes on a monitor anyway.

[DanK]

Hi Colin,

I am no expert on this, but it seems to me that two different issues are mixed up here. The first issue is what dynamic range is captured. If an image has a greater dynamic range than the sensor can record, then the tonal range is censored at one end or the other (or both). Pixels that are pure white or pure black should be something else. The only solutions are reducing the tonal range (like you do with a GND) or using multiple images (HDR, exposure fusion).

The second issue is mapping one tonal range to another, e.g., with a tone curve. You can take an image and compress its dynamic range to fit an output device with a smaller range without censoring the data. This is analogous to what you are doing with a GND. You would compress the scale and therefore might lose the ability to make fine distinctions, but you would not lose the ends of the distribution.

Am I right so far? If so, I think the OP has a legit concern, even if in the end the image may have to be compressed into a smaller tonal range. I find that when doing landscapes, I often have to extend the dynamic range of my camera (a 50D) by 2 stops, and occasionally 4, to maintain detail at the ends of the distribution. I don't use GNDs because I generally don't have a flat horizon. Instead, I enfuse 2-3 images.

Dan

[vladimir]

2. My second question is somewhat related to the first. Why is it that in 16-bit mode Photoshop still displays 8-bit figures in the info window? One would expect that after conversion the whites would be displayed as 65535 for all the channels, but instead all the channels still show 255. I suspect this is because of limitations of the display and the human eye, but I really am in the dark here (no pun intended).

What version of Photoshop are you using?
In CS4 you can have your info in 8, 16 or 32 bit values.
From here (and from personal experience ):
"Depending on the option you specify, the Info panel displays 8‑bit, 16‑bit, or 32‑bit values"

[MrB]

My personal opinion of snapsort is "avoid it".
It is riddled with inaccuracies and mis-leading/irrelevant statistics ...

Thank you for this alert, Dave.

Philip

[Colin Southern]

Hi Colin,

I am no expert on this, but it seems to me that two different issues are mixed up here.

Hi Dan,

You're not wrong - it's just that most of the time folks just don't need more than 4 to 6 stops of DR, although of course that does depend on the scene. Case in point ... bride in white dress standing next to groom in black suit; DR required? Only 4 stops (so up to 8 wasted). Perhaps that scene has some detail in the shadows that one wishes to reveal ... so perhaps another couple of stops needed there (6 used out of 12). Perhaps the photographer under-exposed by a stop - so now were' down to only 5 remaining (safety margin). So that was pretty much the point I was trying to make; for photography involving only reflected light, folks generally only use 1/2 dozen stops of DR - at which point having a camera with 10 to 14 (which would currently represent pretty much 100% of the DSLR market I would think) - really doesn't make any significant difference to the end image.

On the other hand, if you're capturing incident light (ie "shooting into the light source"), but still trying to protect foreground shadow detail (eg in this photo) then yes - definately - scenes like that can push a camera's DR to the absolute limit - however - it's also worth mentioning that in that situation it's VITAL that the exposure push the sensor right up to the saturation point; most automatic exposures leave 1 to 2 stops of "headroom" that you just can't afford for this kind of exposure).

So I think we're on the same page here - just all depends on the DR of the scene being shot; with the noted exception, most scenes don't push a camera's DR capacity that hard; a lot of DR is discarded - it's just that most folks don't realise what's happening.

[tclune]

So I think we're on the same page here - just all depends on the DR of the scene being shot; with the noted exception, most scenes don't push a camera's DR capacity that hard; a lot of DR is discarded - it's just that most folks don't realise what's happening.

I think it's very important to emphasize that what you say here applies ONLY to the base ISO of your camera. My camera, for example, has 12 EV dynamic range at 200 ISO, but that drops to something like 7 EV by the time you get to 6400.

[Colin Southern]

I think it's very important to emphasize that what you say here applies ONLY to the base ISO of your camera. My camera, for example, has 12 EV dynamic range at 200 ISO, but that drops to something like 7 EV by the time you get to 6400.

Yes - I mentioned that in post #5.

Essentially, one loses 1 stop of DR for each doubling of the ISO (because the physical sensor characteristics aren't changing -- we only amplify the smaller and smaller signal more and more -- and noise floor doesn't move - so in essence the signal range gets less and less which essentially corresponds to a DR decrease.

[jenne1978]

@ all you guys: Thanks! This exactly what I was looking for, and more. Dynamic range is something not many photographers seem to be knowledgable about, so this is the first time I see some expert opinions. Thanks for the reviewsites, they are most helpful in finding out...stuff.

@Vladimir: Photoshop seems to be full of these "hidden" buttons, thanks! One more mystery solved.

@Colin: Thanks for the dynamic range info, sometimes a bit of a rant but I must admit being a Photoshop teacher it sounds familiar I always seem to push my equipment, never completely satisfied with sharpness, dynamic range etc. If I only could afford an expensive digital back for my old Fuji 680...
I often do need more than the dynamic range offered by my puny Canon 30D for making interior, architecture and landscape type of shots. Anyways, the DR backstory about it not mattering much in most cases holds true to a lot of other myths about equipment and puts things in perspective I think, proving the fact that I am able to make awesome pictures with pretty high dynamic range and resolution with a DSLR that is considered ancient and low res these days!

[jenne1978]

@tclune:
I finally found the measurements tab. I managed to select a camera that didn't have any measuremnts, so it took a while.
So, let me get this straight. Some Phase One has just a few stops more dynamic range than a Canon 5D Mark II, and that is probably the result of low noise, so an indoor scene with a bright window in the background will hardly benefit from it, am I right? I was told by a Hasselblad owner that he thought the dynamic range is great, but that seems to be a bit overblown now.
Cheers,
Jenne Hoekstra

[tclune]

So, let me get this straight. Some Phase One has just a few stops more dynamic range than a Canon 5D Mark II, and that is probably the result of low noise, so an indoor scene with a bright window in the background will hardly benefit from it, am I right? I was told by a Hasselblad owner that he thought the dynamic range is great, but that seems to be a bit overblown now.

High dynamic range is not a property of either end of the spectrum, but of the entire exposure range. Personally, I value high dynamic range a lot more than Colin does. Of course, he's a massively better photographer than I am, so this may just show that a dull-normal needs more forgiving equipment than someone who knows what he's doing. And I suspect that anyone getting the Phase One would be more interested in the high pixel count than in the high dynamic range (which you can get on a current-generation Nikon for a whole lot less money), but high dynamic range is (as Martha Stewart would say) a very good thing. You need to do your own tone mapping in post to take advantage of it, but that is not a big deal. If you have never wished that you could do an HDR to get the details in a scene at high and low illumination, then you have no need for high DR. But, if you often find that you want to pull detail out of the shadows as well as detail out of the highlights, then higher dynamic range is what you need. As with HDR, it is easy to turn an interesting photo into a grotesque cartoon if you overdo it. But, with careful attention to detail, you can pull out the detail you want in a way that seems very natural without having to contend with the multiple exposures fusion of multiple images.

[jenne1978]

I think we are on the same page here, thanks!

[Colin Southern]

Personally, I value high dynamic range a lot more than Colin does.

Hi Tom,

In my opinion the higher the DR capability of the camera the better (for the greater shadow-detail options from a low-ISO shot, and the higher ISO modes that a high DR equates to at the other end), but personally, I'm finding that it's just not something I struggle with often. In reality, there's only a finite number of things that one can do (without resorting to multiple capture techniques) ...

1. Don't waste DR by under-exposing - which means shooting RAW and learning just how far you can push the exposure before clipping occurs (and by clipping I mean REAL clipping - not the apparant clipping that may be shown on the ultra-conservative in-camera-generated JPEG). In my experience, a LOT of folks who thing they have a DR problem are under-exposing up to 2 stops, which is crazy because it's throwing away a full 3/4 of the information captured (and that's the BEST / cleanest information) (it's like throwing away the meat and keeping the bones).

2. Use a GND filter (even with uneven transition zones (excuse the pun!) they can still be quite effective).

3. Provide additional lighting (essential for interior architecture shots because the most flattering shots will come from pleasing local contrast, and very seldom by simply capturing and revealing existing lighting)

Whether the problem is lack of DR or poor technique (or a combination of both) it's hard to say - but it would be interesting to look at any RAW files that folks would care to offer to see how well they've nailed their exposures.