Canon 5d Mark III Shadow recovery

[Aglet]

I am SO enjoying this.

and now you can finally see the useful relevance of actually shooting with a lens cap on and pushing a mere 4 stops to compare N & FPN. (see my tech blog if you forgot) It sure showed me a lot more useful and easily compared info than all this volleyed text with "real world images."

@Trumpet
so the 5d3 is actually showing reasonable levels of FPN, by which I mean, it's fairly acceptable and workable. That's good. I still won't buy one tho, I want more of an improvement in low ISO performance. After all, I've seen "the dark side," bought it, shoot with it, bought more of it, and I won't go back to Canon's chroma-snow and stripes.

OTOH, the 5d2 I had was SO bad that shades a mere 3EV below metered 0, pushed a mere ONE stop, showed FPN.
And all the handwringing angst THAT generated from so many posters here.
Other 5d2s certainly looked to perform better than the one I had, but it was pretty near impossible for me to have my point accepted and nobody d/cared to supply comparison shots.

People are right, this IS getting boring, all the more so when discussions degrade into little more than semantics.

But I'm glad you posted some garden shed shadow recovery from your 5d3, it's way the heck better than my 5d2 was in that regard.
And you should borrow a d800 and reshoot that garden shed the same way and push the files the same way.
Then report back to us on that.
I suspect I know what that'll be but at least you may then have a bit more respect for what that Exmor can do and why some of us prefer the expanded lower limits it provides.

 

[jrista]

Okay, so maybe my SIWOTI is a bit worse than I thought -- or maybe I'm just dragging my feet a bit before I get back to trying to figure out how I'm going to light this giclee shoot of a beetle that's mostly a solid block of gold leaf the size of a soup bowl.

But here's one more -- perhaps the ultimate -- variation on the theme.

This is the darkest exposure i could get without resorting to filters: 1/8000 @ f/22 @ ISO 100. As far as the back-of-the-LCD preview goes, i might as well have left my lens cap on -- which, I know, is another favorite activity of Internet measurebators. It worked out to nearly ten stops of digital push; that's the digital equivalent of ISO 51,200. As in, take all of the analog gain that the camera does on the sensor at its native maximum ISO of 25,600, do that digitally instead, and then add yet another stop of digital gain.

I'd show you the as-is shot, but it's as solid black as the LCD preview was.

But I am including a 100% crop of the highlights. Yes, the shadows (and midtones) in this shot are junk. Duh! It's a digitally-pushed ISO 51,200 equivalent exposure. But these highlights are what you get from the 5DIII when you push your below Zone I solid featureless black ink blobs to Zone VIII (edit: forgot an "I") and IX near-paper-white highlights.

And if any of all y'all really want -- or think you need -- more out of a camera...well, sorry, but I ain't got no sympathy for you. None whatsoever.

Cheers,

b&

I think the point everyone is trying to make is...those "Zone I solid featureless black ink blobs" are NOT FEATURELESS!! There is obviously detail there. Your latest shots demonstrate the problem Canon sensors have quite well (and the fact you had to pull 10 stops means you are not in a DR-limited scenario, so your previous examples did not have deep enough shadows in the first place.) Do the EXACT same test with a D800. I think your mind will be blown. I can't say any more, because obviously a thousand words isn't enough. You'll have to see for yourself. Borrow, rent, buy, whatever...but if you do the exact same experiment...1/8000th f/22 ISO 100 with a D800, and lift...you won't see any pattern or banding at all. You MIGHT see a little bit of random grain...maybe. The rest...all of those "featureless black ink blobs"...well, they won't be featureless, they won't be black, and they won't be blobs.

Enjoy being knocked on your ass by amazing DR!

 

[TrumpetPower!]

Could you post some original images with full EXIF? The original "dark" image looks a little odd...too flat, such that it makes me wonder if it really is a natively under-exposed shot.

Additionally, the noise in those shadows is not clean.

So, you think I faked the shot but wasn't smart enough to fake it without noise? Really? Or are you just trolling now?

I don't have anywhere to upload RAWs to, but here's a screenshot. If that's not enough for you, feel free to think I'm a liar if that'll make you feel better about your D800.

It looks nothing like the kind of deep NOISELESS or NEAR-NOISELESS shadows you get out of a D800 when it is lifted 6 stops.

So your minivan with the racing stripes doesn't kick in the rev limiter until 105, but my boring minivan's limiter keeps me under 95.

Yawn. Whatever. Good for you. Vroom! Vroom! Yay.

b&

 

[TrumpetPower!]

I am SO enjoying this.
And you should borrow a d800 and reshoot that garden shed the same way and push the files the same way.

Why?

I mean, really.

That vent in the 100% crop of the pushed-ten-stops shot has very little noise left in it, even if it's a bit more textured than sharp.

I took something that would normally come out as solid black, turned it into a Zone IX textured highlight, and it's more than adequate for even significant enlargements.

If you can do a bit of math, that means that the 5DIII has, effectively, at least twenty stops of usable dynamic range: ten from the normal highlights these would have been had it been properly exposed to the solid blacks of the standard rendering of this scene, and then another ten from the digital push of those blacks back to highlights.

And, realistically, for me, it actually works out to thirty stops of usable dynamic range. I'd never want to push something from solid black to a highlight, though I may well want to push something as dark as this to a Zone II textured black. -- and the bottom shadows of this ten-stops-underexposed shot would make a fantastic Zone II rendition, even though they're thirty stops darker than the highlights I would have captured with a normal rendition.

That is, imagine a scene like this but with some sort of coal mine entrance in it. If there was a grating like this in said coal mine and the proper exposure there was thirty stops below the outside-the-mine scene, I could render said grating as a readily identifiable -- though very dark -- grating, and not have any visible noise in said rendition.

...and I should care that a D800 has even more dynamic range...why, exactly? So I could render the black grating in the coal mine with as much texture as I could render a white grating in a coal mine?

Really?

No, really. When on Earth am I going to be shooting in a scene with more than thirty stops of dynamic range that I can't just as easily bracket -- assuming I can't fix the light in the first place?

Again, I really couldn't care what insane speed the pickup's rev limiter is set at, so long as I can reliably and comfortably haul a load of sod with enough oompf left over to pass the semi up the hill.

b&

 

[jrista]

Could you post some original images with full EXIF? The original "dark" image looks a little odd...too flat, such that it makes me wonder if it really is a natively under-exposed shot.

Additionally, the noise in those shadows is not clean.

So, you think I faked the shot but wasn't smart enough to fake it without noise? Really? Or are you just trolling now?

I don't have anywhere to upload RAWs to, but here's a screenshot. If that's not enough for you, feel free to think I'm a liar if that'll make you feel better about your D800.

It looks nothing like the kind of deep NOISELESS or NEAR-NOISELESS shadows you get out of a D800 when it is lifted 6 stops.

So your minivan with the racing stripes doesn't kick in the rev limiter until 105, but my boring minivan's limiter keeps me under 95.

Yawn. Whatever. Good for you. Vroom! Vroom! Yay.

b&

Well, I think your ratios are off there. Remember, as top is a factor of two. If 95mph is the equivalent of 12 stops, then the D800 would be like driving at 380mph!

The D800's edge is TWO STOPS...or TWO FACTORS OF TWO better than the 5D III. We aren't talking a marginal difference, it's measurable and meaningful....at least at ISO 100. Again, we are not saying the 5D III is a bad camera. It most certainly is not, and it obviously takes phenomenal photos. I plan on getting one myself if the 7D II doesn't make some big waves. We aren't bashing on the 5D III...just trying to point out that in certain circumstances, the D800 can be WAY better. Two factors of two better...not 105mph...380mph!

It is also important to point out that the 5D III is better in a number of other ways. At higher ISO, it definitely has the edge. I've seen some amazingly clean ISO 12800 shots from the 5D III that look better than my 7D's ISO 3200. The AF system practically makes me giggle like a little girl, and combined with 6fps, it is an excellent camera for what I do, which is birds and wildlife most of the time (which means I'm always at high ISO and need high frame rates, two things the D800 is NOT good at by any measure.)

 

[TrumpetPower!]

Well, I think your ratios are off there. Remember, as top is a factor of two. If 95mph is the equivalent of 12 stops, then the D800 would be like driving at 380mph!

The D800's edge is TWO STOPS...or TWO FACTORS OF TWO better than the 5D III. We aren't talking a marginal difference, it's measurable and meaningful....at least at ISO 100. Again, we are not saying the 5D III is a bad camera. It most certainly is not, and it obviously takes phenomenal photos. I plan on getting one myself if the 7D II doesn't make some big waves. We aren't bashing on the 5D III...just trying to point out that in certain circumstances, the D800 can be WAY better. Two factors of two better...not 105mph...380mph!

Spinal Tap - 11

Cheers,

b&

 

[jrista]

I am SO enjoying this.
And you should borrow a d800 and reshoot that garden shed the same way and push the files the same way.

Why?

I mean, really.

That vent in the 100% crop of the pushed-ten-stops shot has very little noise left in it, even if it's a bit more textured than sharp.

I took something that would normally come out as solid black, turned it into a Zone IX textured highlight, and it's more than adequate for even significant enlargements.

If you can do a bit of math, that means that the 5DIII has, effectively, at least twenty stops of usable dynamic range: ten from the normal highlights these would have been had it been properly exposed to the solid blacks of the standard rendering of this scene, and then another ten from the digital push of those blacks back to highlights.

And, realistically, for me, it actually works out to thirty stops of usable dynamic range. I'd never want to push something from solid black to a highlight, though I may well want to push something as dark as this to a Zone II textured black. -- and the bottom shadows of this ten-stops-underexposed shot would make a fantastic Zone II rendition, even though they're thirty stops darker than the highlights I would have captured with a normal rendition.

That is, imagine a scene like this but with some sort of coal mine entrance in it. If there was a grating like this in said coal mine and the proper exposure there was thirty stops below the outside-the-mine scene, I could render said grating as a readily identifiable -- though very dark -- grating, and not have any visible noise in said rendition.

...and I should care that a D800 has even more dynamic range...why, exactly? So I could render the black grating in the coal mine with as much texture as I could render a white grating in a coal mine?

Really?

No, really. When on Earth am I going to be shooting in a scene with more than thirty stops of dynamic range that I can't just as easily bracket -- assuming I can't fix the light in the first place?

Again, I really couldn't care what insane speed the pickup's rev limiter is set at, so long as I can reliably and comfortably haul a load of sod with enough oompf left over to pass the semi up the hill.

b&

You don't need 30 stops. According to DXO screen DR, which is the direct measure of the native capabilities of a sensor, the 5D III has 10.97 stops of dynamic range. If you try to photograph a scene with 12 stops, your in a DR-limited scenario. Your then faced with a choice...clip the highlights, or block the shadows. Your clipping a whole STOP...that is, DOUBLE the range of light your camera is capable of sensing at once. This is an extremely common occurrence in landscape photography, where dynamic range, especially around sunrise or sunset, can easily surpass 11 stops, and are sometimes 20 stops or more!

Historically, photographers have used graduated neutral density filters to tackle this problem. Drop in a couple of GND filters of varying transitions and strengths, and you can control the dynamic range of the light reaching the sensor. Sometimes you need to add four, five, six stops of above-horizon filtration to allow you to capture the scene. Stacking GND filters introduces more layers of optical material, which has a degrading impact on IQ, and isn't particularly desirable.

The D800 would handle your 12-stop scene no problem. It'll even handle a 13.2 stop scene and allow you to recover right down into the deepest level shadows without any fpn or banding noise at all, and only a hint of grain.

I'll be happy when Canon releases a 12 or 13 stop sensor. I'll be even more happy when Canon, Sony, or Nikon releases a new sensor with a 16-bit ADC, as that opens up ANOTHER two full stops...another two DOUBLINGS, of sensitivity range with DSLRs.

 

[jrista]

Well, I think your ratios are off there. Remember, as top is a factor of two. If 95mph is the equivalent of 12 stops, then the D800 would be like driving at 380mph!

The D800's edge is TWO STOPS...or TWO FACTORS OF TWO better than the 5D III. We aren't talking a marginal difference, it's measurable and meaningful....at least at ISO 100. Again, we are not saying the 5D III is a bad camera. It most certainly is not, and it obviously takes phenomenal photos. I plan on getting one myself if the 7D II doesn't make some big waves. We aren't bashing on the 5D III...just trying to point out that in certain circumstances, the D800 can be WAY better. Two factors of two better...not 105mph...380mph!

Spinal Tap - 11

Cheers,

b&

Heh, we aren't just talking about finer gradations, which is what the whole "Spinal Tap Crank it up to 11" joke would be in this context. That is bit depth. Increased bit depth allows finer gradations, finer measurement, smaller discrete measurements. Regardless of what the dynamic range is.

Dynamic range is about sensitivity range. That doesn't have to do with fineness...it has to do with capability. Better dynamic range increases the capabilities of a camera. A higher bit depth improves the fineness of how its capabilities are differentiated. This isn't like "Cranking it up to 11", its more like "Using 8 speakers instead of 2".

 

[TrumpetPower!]

According to DXO screen DR, which is the direct measure of the native capabilities of a sensor, the 5D III has 10.97 stops of dynamic range. If you try to photograph a scene with 12 stops, your in a DR-limited scenario. Your then faced with a choice...clip the highlights, or block the shadows.

So, clearly, if I was to take a scene with some solid black areas in it and push those solid blacks ten stops, they'd still be solid black because the camera only has twelve stops of dynamic range and those solid blacks would still be clipped.

...except, of course, that the exact opposite is the case, if you've been paying even the slightest bit of attention to what I've been posting in this thread.

So, one or more of three possibilities presents itself.

I could have been faking everything; the 5DIII might actually have more than twelve stops of usable dynamic range; and / or you might not have a very firm grasp on what dynamic range actually is.

I'm pretty sure I didn't fake anything, so I'm going to go for both of the other two options.

b&

 

[jrista]

According to DXO screen DR, which is the direct measure of the native capabilities of a sensor, the 5D III has 10.97 stops of dynamic range. If you try to photograph a scene with 12 stops, your in a DR-limited scenario. Your then faced with a choice...clip the highlights, or block the shadows.

So, clearly, if I was to take a scene with some solid black areas in it and push those solid blacks ten stops, they'd still be solid black because the camera only has twelve stops of dynamic range and those solid blacks would still be clipped.

...except, of course, that the exact opposite is the case, if you've been paying even the slightest bit of attention to what I've been posting in this thread.

So, one or more of three possibilities presents itself.

I could have been faking everything; the 5DIII might actually have more than twelve stops of usable dynamic range; and / or you might not have a very firm grasp on what dynamic range actually is.

I'm pretty sure I didn't fake anything, so I'm going to go for both of the other two options.

b&

There is no such thing as "solid blacks". That is what I mean about applying an archaic system like the Zone System to the digital era. Solid black doesn't exist. In the scene, its just detail with lower illuminance. The only time you would have a true "solid black" is if you photographed something that was solid black in a closed, sealed room with no light. Or photographed a cave without any flash. In those cases...you have no light at all. That is the SOLE means of achieving this theoretical "solid black".

Any other time, there is going to be light. If there is any light in a scene, some of it will reach your sensor. Those deep shadows, when photographed with a 5D III won't actually be black. Lift the shadows a few stops, and you'll see noise. Those same deep shadows when photographed with a D800 would be deep shadows, with detail that can be recovered.

The difference is that while the 5D III sensor is probably capable of registering most of the light from those deep shadows in your scene, and IS actually probably recording some detail, when the analog signal is converted to a digital one by Canon's off-die ADC's in the DIGIC 5 chip, that detail is overridden by noise introduced into the signal by the ADC electronics themselves. THAT is where all that nasty, noisy detail comes from, and THAT is what reduces the 5D III's dynamic range from whatever the sensor itself is probably capable of, which is probably similar to the D800's Exmor sensor, to the 10.97 stops that you actually get. Dyanmic range. The ratio between maximum signal and the noise floor. NOISE FLOOR. Without the noise floor, or with a much lower noise floor, you'll have more dynamic range. The D800 doesn't introduce as much noise, hence the greater dynamic range...and the reason why "solid black" is actually "deep shadows with detail" rather than "deep shadows with noise and muddy, useless ink blobs".

 

[art_d]

TrumpetPower, I don’t understand what you are trying to show. That pushing a Canon raw file six stops makes it look terrible? We already knew that.

There have been actual examples posted showing where more native dynamic range would be useful. Do you think posting examples where it is less useful changes anything? I have hundreds of shots myself where DR is not an issue. But that doesn’t change what I showed in my examples.

Maybe it’s just me but it comes across as sidetracking the discussion. Trying to make the point go away by directing attention to something else. Let’s recall the following:

- You asked for such real world examples. I provided them. You proceeded to try to discredit them (and by proxy my competence as a professional photographer, though perhaps the latter was unintentional).

- You seem to accept that GND filters (and by extension exposure blending) are valid tools for light modification. But when pointed out that these tools exist to compensate for dynamic range limitations, you made an argument that prints have not been affected by the dynamic range limitations of cameras for decades. I explained this is not the case, again referred to an example of my own print, and observed that many other photographers have (and continue) to use GND filters and exposure blending for printed work to compensate for DR limits. You seemed to ignore this point.

- You also seem to have ignored my point that both content and context are important factors with prints. So any judgement of what is “acceptable” without a discussion of its context is incomplete.

The impression that comes across is you are not actually interested in evaluating the practical limits of dynamic range and where it can impact photography. I understand DR is not an issue for you, and that’s fine. But the logic of “I don’t have issues with DR so therefore no one should have issues with DR” doesn’t really work.

Still, I hope that this discussion has been useful for others who might read it.

 

[BrettS]

Say, that's a real nice shed.

 

[TrumpetPower!]

TrumpetPower, I don’t understand what you are trying to show. That pushing a Canon raw file six stops makes it look terrible? We already knew that.

If you think that the pushed-six-stops image here:

http://www.canonrumors.com/forum/index.php?topic=10008.msg253382#msg253382

looks terrible, then either I'm truly blind or you're not living in the real world.

Either way, we've clearly reached the end of this discussion.

Or perhaps you didn't actually read the text accompanying said post? Maybe you're under the mistraken impression that I only pushed the shadows in said image? You do realize that I pushed the entire image six stops, no?

Because, to my eye, at least, everything from Zone IV and up is perfectly free of noise and as sharp as anything you'll see. Zone III is starting to get a tad gritty, but it's nothing you couldn't easily clean up -- and it certainly wouldn't show even if you didn't clean it up in a 12" x 18" print. And the noise in the shadows darker than that looks like the texture of the objects themselves, so it's a non-factor.

And, again. Every single pixel in that image has been pushed six stops.

I'd call that freakin' amazing, not terrible.

But maybe I'm blind, or maybe you're insane.

Either way, best of luck with your clearly inadequate Canon gear. Maybe it's not too late to trade it in for a Nikon that you'll be so much happier with?

Cheers,

b&

 

[art_d]

TrumpetPower, I don’t understand what you are trying to show. That pushing a Canon raw file six stops makes it look terrible? We already knew that.

If you think that the pushed-six-stops image here:

http://www.canonrumors.com/forum/index.php?topic=10008.msg253382#msg253382

looks terrible, then either I'm truly blind or you're not living in the real world.

Either way, we've clearly reached the end of this discussion.

Or perhaps you didn't actually read the text accompanying said post? Maybe you're under the mistraken impression that I only pushed the shadows in said image? You do realize that I pushed the entire image six stops, no?

Because, to my eye, at least, everything from Zone IV and up is perfectly free of noise and as sharp as anything you'll see. Zone III is starting to get a tad gritty, but it's nothing you couldn't easily clean up -- and it certainly wouldn't show even if you didn't clean it up in a 12" x 18" print. And the noise in the shadows darker than that looks like the texture of the objects themselves, so it's a non-factor.

And, again. Every single pixel in that image has been pushed six stops.

I'd call that freakin' amazing, not terrible.

But maybe I'm blind, or maybe you're insane.

Either way, best of luck with your clearly inadequate Canon gear. Maybe it's not too late to trade it in for a Nikon that you'll be so much happier with?

Please accept my apologies and in place of where I said "six stops" note that I meant "ten stops."

All of my points which I made in that post, however (which you did not comment on) still stand regardless.

 

[TrumpetPower!]

The rest of my points, however still stand regardless.

Apology accepted...but now I'm really confused.

If we agree that there's no trouble cleanly pushing an image six stops, maybe more...well, how is that not enough for all but the most extreme photographic situations?

I mean, we're talking about easily photographing scenes with at least sixteen stops of dynamic range from brightest highlight to darkest shadow, and retaining noise-free detail in everything.

Are you really trying to tell me that it's not an extreme situation to need even more than that?

And you still seem to not wish to address the matter of your pixel-peeping clients that you're so concerned about, and instead keep going off on tangents about consistency of presentation.

Presentation of what? Reasonable-sized prints, or pixel peeping and huge enlargements?

Do your clients not actually pixel peep? Do they pixel peep but nobody actually cares about what they see when they do so? Or are you somehow not quite understanding that there's no difference between pixel peeping and shoving your nose in a door-sized print?

Cheers,

b&

 

[jrista]

The rest of my points, however still stand regardless.

Apology accepted...but now I'm really confused.

If we agree that there's no trouble cleanly pushing an image six stops, maybe more...well, how is that not enough for all but the most extreme photographic situations?

I mean, we're talking about easily photographing scenes with at least sixteen stops of dynamic range from brightest highlight to darkest shadow, and retaining noise-free detail in everything.

I think you are misunderstanding what photographic dynamic range is. No camera is currently able to photograph sixteen stops of dynamic range in a single photo. According to the statistics, we'll use DXO since that seems to be the only source people really trust. According to the Screen DR setting, which is a direct measure without any processing...the 5D III has 10.97 stops of DR, while the D800 has 13.23 stops of DR. Neither of those cameras are able to photograph a scene with 16 stops of DR, and reproduce all 16 on screen.

For one, the dynamic range of the output image is limited by the bit depth of the ADC, and since both cameras are 14-bit cameras, the maximum native DR is theoretically 14 stops, regardless. The D800 has very low read noise thanks to the design of Exmor and its low frequency digital readout and CP-ADC, and it loses only about 2/3rds of a stop to noise (hence the 13.23 stops of DR). The 5D III has low read noise, but it does not have the benefit of on-die CP-ADC or digital readout, so the analog signal picks up noise when it's shipped over the bus to the DIGIC 5 DSP, and the high frequency ADCs inject a whole lot more noise. That reduces the 5D III's DR by a full two stops!

If you photograph a scene with 16 stops of dynamic range, neither the 5D III nor the D800 is going to be able to capture the scene without either clipping highlights, or losing detail in the shadows to noise. The difference is that the D800 only loses two stops, while the 5D III loses four stops. Stops are factor-of-two...if we reduce the difference between the D800 and 5D III in terms of capturing a 16 stop scene in a single frame to a scalar score where lower is better, the D800 gets a score of 4, while the 5D III gets a score of 16!

Are you really trying to tell me that it's not an extreme situation to need even more than that?

Is there a context within which you wish to apply that statement? You seem to be ignoring a very obvious example where photographers will probably always need and be able to use more dynamic range.

Have you ever photographed landscapes? When it comes to landscape photography, the need to photograph scenes with 14, 16, 20 stops is not uncommon, on the contrary it is quite common. It's why every serious landscape photographer treks around with a stack of GND filters...we ALWAYS face DR problems, and the only way to mitigate them is either by optically changing the scene dynamic range with filtration...or with a camera that can natively capture more in the first place, thus improving the quality of shadow recovery.

GND filters are less than ideal. They can reduce IQ, and in scenes with truly extreme DR, stacking strong GND filters is often necessary to reduce the sky enough to avoid clipping. That usually results in unnatural shading on any landscape or scenery that peeks above the horizon...such as mountains, trees, etc....producing a less than ideal shot, or a shot that does not fully replicate the artists vision. A camera capable of more dynamic range means you have to use less filtration, or may even allow you to ditch filtration altogether and just capture the scene as it really is, and fully realize your artistic goals.

 

[TrumpetPower!]

No camera is currently able to photograph sixteen stops of dynamic range in a single photo.

And yet, I've posted an example of exactly that in this very thread. Huh. Fancy that.

Indeed, not only have I posted an example of capturing sixteen stops in a single exposure, I've posted an example of how one could go about capturing thirty twenty stops in a single exposure, though, granted, the bottom end would be a bit on the muddy side -- but still quite usable if you know what you're doing with it.

Maybe I'm not the one who doesn't understand what dynamic range is...?

Pro tip: don't tell somebody something can't be done after he's just done it.

Neither of those cameras are able to photograph a scene with 16 stops of DR, and reproduce all 16 on screen.

That would be quite a feat, indeed, considering the ten stops of the ICC PCS that applies to every RGB image in any color-managed environment. Not to mention how uncomfortably bright -- painful, even -- a display with that much dynamic range would have to be.

Of course, if you understood what dynamic range is, you'd understand all that, too.

Have you ever photographed landscapes?

I see you haven't paid the slightest bit of attention to a single thing I've posted in this thread, for I've included at least three landscape photographs of mine, including one that captures detail not only from the shadows at the bottom of the Grand Canyon but in the very Sun itself, all in the same image.

You, on the other hand, seem to be much more interested in waxing poetic about DXOs and Exmors than actually taking pictures.

Cheers,

b&

 

[art_d]

The rest of my points, however still stand regardless.

Apology accepted...but now I'm really confused.

If we agree that there's no trouble cleanly pushing an image six stops, maybe more...well, how is that not enough for all but the most extreme photographic situations?

We do not agree on that. Because pattern noise will be more clearly evident in some photos more than others. (As an example I refer again to the 20x30 inch print).

The content of the photo does matter. I could post examples all day long of photos where pattern noise isn't an issue. So what? That does not invalidate the examples of when it does happen. Such as the two examples I provided.

I mean, we're talking about easily photographing scenes with at least sixteen stops of dynamic range from brightest highlight to darkest shadow, and retaining noise-free detail in everything.

Are you really trying to tell me that it's not an extreme situation to need even more than that?

jrista already answered this above. I will just add that I do not believe "extreme" is the correct term to describe the photos I have presented. This is probably an issue of semantics. However, as has been repeated, valid tools such as GND filters and exposure blending exist for a reason: to deal with DR limitations. The very existence and common use of those tools should be proof enough that the number of situations where DR limitations happen are not as negibible as the phrase "extreme situation" would seem to imply.

And you still seem to not wish to address the matter of your pixel-peeping clients that you're so concerned about, and instead keep going off on tangents about consistency of presentation.

Presentation of what? Reasonable-sized prints, or pixel peeping and huge enlargements?

Do your clients not actually pixel peep? Do they pixel peep but nobody actually cares about what they see when they do so? Or are you

I think you are muddling the discussion. Let me make it clearer:

a) I talked about consistency of quality in my image files that I provide to my clients. This means that any given image file will be perceived as having roughly the same level of quality as any other file. Scenes will all be exposed correctly, images have a certain minimum level of resolution, and I will not give a client any files which contain fixed pattern noise. You made the suggestion that in the prison door photo, any noise would be marginalized by smaller print sizes, and who would want to print that photo bigger? Well, it's not my place to say. I can tell you that I've been surprised a few times by which images people choose to make prints of. So I won't presume to tell them what images they can have at what print size. And as I've already said, fixed pattern noise can be an issue in even moderately sized prints.

b) Separately, I talked about the context in which a print is presented and how that affects what one might consider "acceptable." Again as an example, the amount of noise or detail smudging that I would tolerate in a print for display in a conference room is different from what I would tolerate for a gallery print.

And as I've explained, I have seen pattern noise become objectionable in reasonable-sized prints. You probably would not see it you were standing five feet away. But you would if you standing a two feet away. And I am not about to rope off a five-foot area in front of my prints so people can't get too close to them.

 

[Aglet]

I am SO enjoying this.
And you should borrow a d800 and reshoot that garden shed the same way and push the files the same way.

Why?
I mean, really.

But why should you try this with a D800 or other Exmor-equipped camera?.
Why not?
Call it a learning experience, pushing the/your envelope.

Want an application?
Shooting into the sun and being able to underexposure more to capture color gradients even closer to old Sol and still bringing up the rest of the scene to visible levels and retaining more color and tonal information without FPN.

Edit: here's an example of just that:
www.canonrumors.com/forum/index.php?topic=8105.msg161888#msg161888
Significant PP work to re-tone the original image to get what I wanted from it, so that it matches my perception of how it looked, standing on that beach. No NR was used or needed. There are no pure black pixels. There are no pure white pixels. There is no FPN. (I did remove an ugly boat.) It looks good in print, 36" wide, could go double that easily. /edit

Or, underexposing more to capture textural information on objects with lots of specular highlites or highly reflective areas while also still retaining good liftable data on the rest of the scene.
I actually did such a shot last week with a D5100 as I was poking around a snowy rural landscape in bright sunshine and shot a subject that also contained very deep shadow information and I paid close attention to how it looked visually so I could recreate it later. Come to think of it, coal may actually have been involved!
It's not as good or extreme an example as a sunset, it's not even a great shot but I took it as an exercise to examine later.
That you can't imagine situations where this is useful surprises me and makes me wonder about your range of photographic experience. Some of us like to try extreme things for the sake of it, to discover where the limits are. For some of us, that edge is where the fun and learning happens.

I took something that would normally come out as solid black, turned it into a Zone IX textured highlight, and it's more than adequate for even significant enlargements.

if that's something that would normally come out as solid black I don't want you doing my printing or prep work.
Perhaps you like an overly contrasty tone curve - some do. I certainly don't.

If you can do a bit of math, that means that the 5DIII has, effectively, at least twenty stops of usable dynamic range: ten from the normal highlights these would have been had it been properly exposed to the solid blacks of the standard rendering of this scene, and then another ten from the digital push of those blacks back to highlights.

uhmmmm...
The only relevant math is how much DR can be captured in ONE shot and it's qualified by how one defines the lower limit noise floor; whether that's an average RMS value of the noise, or a more useful Peak to peak level of noise (FPN) becoming visible at 100%. The latter is my more stringent standard.
As it stands, the 5d3 and the d800 are still limited to less DR than many daylite scenes can present.
When challenged with such a scene tho, I'll use a more capable D800, thank-you.

BTW, Thanks for supplying a 7 stop under-exposed 5d3 image on the previous page.
I'm slightly impressed that it recovers as well as it does when bringing it back those 7 stops without the banding being any worse.

 

[jrista]

No camera is currently able to photograph sixteen stops of dynamic range in a single photo.

And yet, I've posted an example of exactly that in this very thread. Huh. Fancy that.

Indeed, not only have I posted an example of capturing sixteen stops in a single exposure, I've posted an example of how one could go about capturing thirty twenty stops in a single exposure, though, granted, the bottom end would be a bit on the muddy side -- but still quite usable if you know what you're doing with it.

Maybe I'm not the one who doesn't understand what dynamic range is...?

Pro tip: don't tell somebody something can't be done after he's just done it.

Neither of those cameras are able to photograph a scene with 16 stops of DR, and reproduce all 16 on screen.

That would be quite a feat, indeed, considering the ten stops of the ICC PCS that applies to every RGB image in any color-managed environment. Not to mention how uncomfortably bright -- painful, even -- a display with that much dynamic range would have to be.

Of course, if you understood what dynamic range is, you'd understand all that, too.

Have you ever photographed landscapes?

I see you haven't paid the slightest bit of attention to a single thing I've posted in this thread, for I've included at least three landscape photographs of mine, including one that captures detail not only from the shadows at the bottom of the Grand Canyon but in the very Sun itself, all in the same image.

You, on the other hand, seem to be much more interested in waxing poetic about DXOs and Exmors than actually taking pictures.

Cheers,

b&

Before I even begin to address the rest of that, you need to explain how the scene of the shed you have used demonstrates SIXTEEN STOPS of dynamic range. You claim it is, you have even claimed its as much as THIRTY stops, but you have not offered any explanation as to what you mean by "stop of dynamic range" or how you arrive at 16 or 30 stops worth, especially from the shed scene (which doesn't even come remotely close to having 10 stops of DR, let alone 16 or 30).

I don't agree with what DXO claims with their Print DR numbers (and anyone who knows me on these forums can attest to that), but I do trust their Screen DR numbers. There isn't a DSLR on the planet that has electronics (sensor + ACD) capable of producing a RAW image with more than 14 stops. Theoretically, it's impossible. The best there is is the D800 going by DXO numbers, and that tops out at 13.23 stops.

Another excellent and reliable source of information on this kind of thing, at least for Canon sensors, is Roger Clark's site ClarkVision.com. Clark's site gets a million hits a month, and he has a Ph.D and has done some prestigious work for NASA involving optics and digital imaging for some of their mars missions. Clark provides extremely detailed analyses, including the data he uses in his analysis, for a number of Canon cameras. The 5D III analysis can be found here, and it clearly states 10.9EV at ISO 100. Also at ISO 100, you can see that read noise (which enjoys a small contribution from the sensor readout itself, from amplification, and primarily from ADC) is at its highest of 34.9 electrons, and at ISO 200 is 18.3 electrons. The increase in read noise is why Canon sensor DR flattens out at ISO 400, 200, and 100 rather than maintaining a linear increase right up to 14 stops.

According to scientific analyses by two well-known and trusted sources, DXOMark and Ph.D Roger Clark, the 5D III doesn't even achieve a full 11 stops of DR. In light of that, I believe it is only fair you explain to everyone here how you have determined the 5D III is capable of capturing a scene with sixteen stops of DR in a single shot, or at least how you are evaluating your shed shot and processing to arrive at either 16 or 30 stops of DR. Because it doesn't conform with any other form of measurements I've ever come across.