Film versus DSLR dynamic range

Started Jul 24, 2007 | Discussions
PIXSurgeon Veteran Member • Posts: 4,009
I guess...

Dave Luttmann wrote:

Do your homework instead of playing around with Imatest.

...it is your homework, Dave, the one in order here.

You simply have shown nothing , except conceptually confining dynamic range to bit depth, which, as you already know, it is innacurate.

What I find funny here is the following:

1. You complain about "IMATEST testers" not being connected to reality.
2. You state that DR (in "real world") has nothing to do with IMATEST.

3. I come and show you ACTUAL images, from the "real world" I suppose, and NOT A SINGLE Imatest chart.
4. You dismiss the evidence with a concept and a computation...

Dave: WTF happened with your "REAL" world images? Can we all see them?

In fact, I propose you the following (so we can provide the audience with some substance, in place of medieval poetry):

1. Send me a strip of the HIGHEST quality 35mm film you may be proud-of, with just few of DR-intensive images, in your "10-11 stops range". I will provide you with the actual place to send.

2. I will use my DiMAGE Scan Elite 16-bit 5400 Scanner (which will pull most of what is usable from your film), and generate the actual digital versions (using my SKILLS, not yours...

3. I will post the results here, for everyone to see, and I will send you a CD.

Let me know if you accept.

8-)))

-- hide signature --

TIP: If you do not like this post, simply press the 'COMPLAINT' button. Mommy/Daddy are just one click away.

ewelch Senior Member • Posts: 1,023
Re: Film versus DSLR dynamic range

It used to be true, but I find my current DSLRs have better dynamic range than film. But I've only shot about 15,000 rolls of film (and processed 99 percent of it myself) so what do I know?

Alfredo Li Pira wrote:

A number of posts in recent threads state that film has more dynamic
range than DSLR sensors. Actually, these measurements

( http://www.clarkvision.com/imagedetail/film.vs.digital.summary1.html ) clearly show the contrary, with slide film having about 5 stops dynamic range, print film about 7 and a DSLR getting to 10 stops. I would be interested in understanding on what are the statements about the low dynamic range of DSLRs based?

Thanks,

ALP

-- hide signature --

Eric

Ernest Hemingway's writing reminds me of the farting of an old horse. - E.B. White

 ewelch's gear list:ewelch's gear list
Canon EOS-1D X
Jay Turberville Forum Pro • Posts: 12,917
Re: Correct

Jay, you're completely correct.

Then maybe you will consider some of my counterpoints. We may generally agree, but apparently we don't totally agree.

First off, with a 12 bit sensor, you cannot physically discern more
than 11 stops.....it's impossible.....period.....end of story.

Well, technically it's the ADC that determines the bit depth...

4096, 2048, 1024, 512, 256, 128, 64, 32, 16, 8, 4, 2, 1, 0

Well, if you had a noise free sensor I think you could go beyond 11 stops to 12. Of course, sensors aren't noise free.

With
a 12 bit system, the 11th stop has only 2 tonal values....on and off.
That is hardly of any use. In fact, even the 8th stop only has 16
tonal values assigned, thus, of little, limited value.

I think that's the 12th stop. But maybe I'm counting wrong. And actually in dark shadows you really don't need a whole lot of tonal steps within a "zone" to get useful texture.

Compared to slide film, DSLRs have decent dynamic range.

Yep.

I have
little problem getting 10 to 11 stops from color neg film and a 16
bit 16X multi scan from Fuji ProS 160 or the old NPS 160. I really
wish people would stop quoting that laughable film test of Clarkes
where he used Kodak 200 consumer grade garbage....known for poor
color rendition and dynamic range.

Sure. I have no good way to test film. I know my scanner is pretty marginal for film and I can't really count on scans from it for meaningful DR info. So I take most reports of film DR with a grain of salt.

With B&W it's a different story altogether. I have routinely
obtained 14 stops with Tri-X using dilute developer solutions and
stand development.

Here's what I find interesting. Ansel Adams' zone system typically makes use of maybe 12 zones at the most. So once you can get around 10 stops of DR, you've covered a huge amount of DR. One of the problems with scenes that have more than 10 stops of DR is managing to get the tones reproduced on a print (with greatly limited DR itself) in an attractive manner.

The problem with digital capture in this respect is its linearity.
Films have a more gentle rolloff of highlights and shadows due to
their response curves at the toe and shoulder. It would require some
nasty overexposure to blow out Tri-X or FP4 compared to a DSLR.

Right. You expose digital to preserve your highlights, and then introduce a tone curve that best fits the sensor's linear output to your output medium. I find that usually means I introduce a toe and shoulder which usually has the result of decreasing shadow and highlight contrast and increasing midtone contrast.

The only people who think DSLRs exceed film in dynamic range are
those who hang out in a lab using Imatest.

Well, I'm an Imatest user and advocate. And I came to my conclusions by my use of Imatest and my experience is shooting digital. Here's a test from my C7070. This camera uses the very same sensor that Clark's Canon S70 uses. Clark says that he gets 11 stops of DR from the sensor at ISO 50. I measured 11.2 stops of total DR with Imatest and a Stouffer T4110 test wedge (no calibrated candles were handy). But the 11.2 measurement is not the meaningful one. The measurement at 0.25 or 0.5 threshold is what experience has taught me is the meaningful one. Beyond those thresholds, noise is usually a problem. My reading of the test is that on a good day with lots of effort, you can get maybe 9 stops of usable DR from a C7070 - even though you can measure 11.2 stops.

Imatest is a fine tool. Ya just need to interpret the results appropriately. BTW, I'd be happy to supply the raw files to anybody who wants to try to show me how there is 11 stops of usable DR in the test image.

In the real world, their
tests fall flat.

Nope. Just the interpretations of them.

I can blather on about lab results for Tri-X that
prove 22 stops of dynamic range....in the real world, it's nowhere
near that.....much in the same way that in the real world, 12bit
DSLRs are nowhere near 11 to 12 stops.

Not practical stops, no.

The wedding photographers I know all realise this through
experience.....not hiding in a lab behind Imatest results.

Oh please, quit picking on Imatest. It gives perfectly fine results that match and predict real world experience if you just use it correctly.

BTW, using it correctly IMO includes using it with raw and trying to avoid the introduction of any tone remapping or noise reduction - as well as making reasonable interpretations of the results.

-- hide signature --
Jay Turberville Forum Pro • Posts: 12,917
Re: I guess...

...it is your homework, Dave, the one in order here.

You simply have shown nothing , except conceptually confining
dynamic range to bit depth, which, as you already know, it is
innacurate.

Actually, until someone starts storing gamma adjusted raw values, it is a perfectly valid representation. Digital sensors are linear in their response and all of them (with the exception of the Fujifilm sensor perhaps) store linear raw values. So that analysis is pretty accurate. It may be off a bit due to things like flair polluting very dark areas though. My tests show a very linear response until you reach those darkest and noisiest tones.

What I find funny here is the following:

1. You complain about "IMATEST testers" not being connected to reality.
2. You state that DR (in "real world") has nothing to do with
IMATEST.

The irony is that Dave and I tend to agree and I happen to think Imatest is a fine tool and use it to come to conclusions that right now seem at odds with yours.

3. I come and show you ACTUAL images, from the "real world" I
suppose, and NOT A SINGLE Imatest chart.

Yeah, but so what? That candle wasn't calibrated to anything. We don't know the brightness values there.

4. You dismiss the evidence with a concept and a computation...

Probably because it is darned poor evidence.

Let me know if you accept.

Hey, maybe I'm wrong and there are 11 stops of usable DR if I can measure it using schemes similar to Clark. I'll be happy to send you raw files of step wedges and you can develope them and show us how there are usable tones beyond 9 or so stops. Every time I go there, the noise gets pretty intense.

-- hide signature --
Jay Turberville Forum Pro • Posts: 12,917
Re: Film versus DSLR dynamic range

It used to be true, but I find my current DSLRs have better dynamic
range than film. But I've only shot about 15,000 rolls of film (and
processed 99 percent of it myself) so what do I know?

Apparently not enough to be more specific about what you mean by "film." The DR of different films varies a lot.

-- hide signature --
PIXSurgeon Veteran Member • Posts: 4,009
And...

Jay Turberville wrote:

Probably because it is darned poor evidence.

...WHERE is yours?

Let me know if you accept.

I'll be happy to send you
raw files of step wedges and you can develope them and show us how
there are usable tones beyond 9 or so stops. Every time I go there,
the noise gets pretty intense.

...I DID NOT ask for Stouffer wedges (I already got them, http://www.pbase.com/feharmat/image/49870707/original , long time ago ).

We are working under Dave's "REAL WORLD" images principle, and I have asked him to step-up to the plate and send me FILM IMAGES, so I can scan them and process them, using my skills and tools, and post the results, here.

Or are you going to live up to Dave's irony's, too?

P.S.: Cut the B.S. and grab your cam, fire your candle, on a low-light area, set the cam to ISO400 and shoot HANDHELD. Come back here, and let's see your "genius" at work...

8-)

-- hide signature --

TIP: If you do not like this post, simply press the 'COMPLAINT' button. Mommy/Daddy are just one click away.

Ravncat Senior Member • Posts: 1,109
Re: Film versus DSLR dynamic range

I remember when digital cameras were a little worse of, or equal to "generic" slide film back in the

Though, I don't understand why people talk about an advantage of film being how it behaves when it overexposes, unless you are going for overexposure, it's not so hard to nail your exposure.

-- hide signature --

A poor photographer blames his tools.

Jay Turberville Forum Pro • Posts: 12,917
Re: And...

Probably because it is darned poor evidence.

...WHERE is yours?

I offered images of step wedges. But that's really beside the point. Whether or not I have offered any evidence has no bearing on the point that your flame image is not a good demonstration of any particular dynamic range capability.

Let me know if you accept.

I'll be happy to send you
raw files of step wedges and you can develope them and show us how
there are usable tones beyond 9 or so stops. Every time I go there,
the noise gets pretty intense.

...I DID NOT ask for Stouffer wedges (I already got them,
http://www.pbase.com/feharmat/image/49870707/original , long time ago
).

I know you didn't. But it naturally follows from your claims that a proper workflow should easily bring 10 stops of DR out of these images. I'm just giving you a chance to prove your point. You, of course, are free to accept the opportunity or decline it.

We are working under Dave's "REAL WORLD" images principle, and I have
asked him to step-up to the plate and send me FILM IMAGES, so I can
scan them and process them, using my skills and tools, and post the
results, here.

The problem with "real world images" is that they have to be metered carefully beforehand if they are going to be used to prove a certain level of DR. That's the problem with your candle image. We have no idea what the brightness values were. So we don't know what dynamic range was captured. And I really don't care what you think "we" are working under. And I'm working under any reasonable approach that can demonstrate actual dynamic range. That can be a "real world" image or any number of reasonable test images or exposure sequences.

P.S.: Cut the B.S. and grab your cam, fire your candle, on a
low-light area, set the cam to ISO400 and shoot HANDHELD. Come back
here, and let's see your "genius" at work...

There's no B.S. to cut. Your evidence just isn't very good is all.

Why handheld? That has nothing to do with dynamic range. Why ISO 400? How is that relevant? The point isn't comparing your camera with mine or to some other camera. The point is whether or not Roger Clark's dynamic range numbers have practical application or not. If you can supply some reasonably good source of info that documents the range of light values that we'd expect from a candle, then I might shoot one since that might make the test releveant. But if we don't have some good information on the brightness levels of the subject, then I don't see how taking a picture of it makes a case for a certain amount of dynamic range. Realizing the problems with shooting something like a candle flame doesn't take genius. It's just obvious.

Anyway, the image below demonstrates my point. It is a pair of nice boring pictures of a Stouffers step wedge. For those who don't know, each step represents a 1/3 stop exposure difference. Two strips are shown. Each is from the same raw file. The bottom one is a a "normal" developement that shows slightly more than 8 stops worth of steps and hence slightly mroe than 8 stops of DR. The darker tones don't show much visible noise. This is the 8 stops or so of DR that I say is comfortably achieved by most DSLRs.

The wedge above that one has been processed to bring the values beyond 8 stops into a useful range. On this image, you can make out steps that go as far as the 11th stop. But there is a lot of noise present in these last two stops. So the range of 11 stops is there, but the question that each photographer has to answer for themself is how useful those last extra stops of DR are? For me, they aren't usually very useful. I'm pretty comfortable pushing into 9 stops, and maybe even 10 in an pinch. - depending on final enlargement size and subject matter. I have no hesitation saying that this camera (E-330) easily has 8 stops of DR. But even though I can measure more, those extra stops don't represent a lot of useful image detail.

http://www.jayandwanda.com/photography/E-330/TT231651_iso100_MaxDR.jpg

Now a camera with a larger pixel pitch and and hence typically greater well capacity will go into the the last three stops of DR more comfortably. But that isn't true of the vast majority of DSLRS. And further, Roger Clark says that even a 7Mp 1/1.8" sensor has 11 stops of DR. And while that that may be true technically, it isn't a good practical preditor of what you can expect when you take pictures. The usable DR is a few stops less. And the typical 6-10Mp sensor on an APS-C or 4/3" camera has a similar problem, just to a lesser degree.

http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html

-- hide signature --
PIXSurgeon Veteran Member • Posts: 4,009
I stopped reading here...

Jay Turberville wrote:

your flame image is not a good demonstration of any
particular dynamic range capability.

...We do not have anything else to discuss.

..

-- hide signature --

TIP: If you do not like this post, simply press the 'COMPLAINT' button. Mommy/Daddy are just one click away.

Mario Senior Member • Posts: 2,648
Bit depth and dynamic range...

You are correct when you describe how many "levels" are available to describe shades of gray in each channel for a 12-bit bitmap image.

However this does not map directly to dynamic range. It is certainly possible to compress the tonal values of the scene that has dynamic range more than 11 stops into this range and still have some measurable detail in both shadows and highlights.

And the compression is usually done at the expanse of mid-tones (i.e lack of contrast). This is why extremely high DR images usually have "sick" looking low contrast.

-- hide signature --

Mario

RomanJohnston
RomanJohnston Forum Pro • Posts: 18,827
I agree.....

Tested my D2X and found about 9.5 stops total (had to do it to calibrate for my Sekonic L-758DR meter)

Now that is with me taking measurments with my eyedropper in photoshop.

For my eye...I saw more like 8.5 stops...that I could actually see.

Roman
--

The Law Of Attraction is ALWAYS working. Your only choice is whether you drive 'it'...or 'it' drives you.
-Me
http://www.pbase.com/romansphotos/

Steve Bingham
Steve Bingham Forum Pro • Posts: 26,479
Back in the "old days" . . .

Many a skilled photographer acquired secrets for extending the DR of film. One was pre-soaking the film in a disproportionate developer and then under developing. A real bitch to print, but the info was there. Of course pre-flashing the film before normal exposure was another. The list goes on. AA (and others)spent many years perfecting these techniques. Hence, the Zone System began and continued on to various printing techniques (the REAL problem child).

Generally speaking Tri-X was known for it's incredible DR and long toe. Again, a bitch to print. I am not sure what a film densitometer would measure, but I would guess 11 stops would be about max. My D200 can do that with a 16 bit double conversion.

Now if we were to compare my favorite slide film of old, Velvia 50 or 100, the D200 wins - by about 4 stops!

For four years I developed about 10 rolls a day, 5 days a week (sports and news). And that's not including MF and sheet film. Then I changed jobs and only did 20 rolls a week (glamor and products). Figure 50,000 rolls over a 40 year period! Now you know why I LOVE DIGITAL!!!! :^) Love it, love it, love it. Don't talk to me about film!
--
Steve Bingham
http://www.dustylens.com
http://www.ghost-town-photography.com

 Steve Bingham's gear list:Steve Bingham's gear list
Nikon D7200 Nikon D810 Nikon AF-S DX Nikkor 18-140mm F3.5-5.6G ED VR Nikon AF-S DX Nikkor 16-80mm F2.8-4E ED VR Nikon AF Nikkor 85mm f/1.8D +24 more
Harry Lime Regular Member • Posts: 264
Re: Film versus DSLR dynamic range

The estimates I have seen give are as follows:

Slide film ~ 5 stops.
DSLR ~ 8 stops (Fuji S5 ~ 10 stops)
Print film ~ 11 stops.

I can see that from my own experience that my D200 has much greater
DR than Fuji Provia 100F and KR64. Assuming I shoot RAW of course.

This mirrors my experience for the past 15 years.

The S5 PRO is very impressive and the fist DSLR I have seen that comes very close to
negative film. Give it another 5 years or and we'll be there.

HL

RomanJohnston
RomanJohnston Forum Pro • Posts: 18,827
Thow in my digital blending technique...

Well...maybe my DR isnt unlimited...but my compression of it into one shot that is managable IS. ; )

HDR still isnt my cup of tea (yet...waiting for it to mature)...but layer masks are stupid simple to make...once you get the hang of it.

I'm with you Steve....give me digital..or give me death....er...um...ahem...you know what I mean...lol!

Roman
--

The Law Of Attraction is ALWAYS working. Your only choice is whether you drive 'it'...or 'it' drives you.
-Me
http://www.pbase.com/romansphotos/

Lictor Regular Member • Posts: 346
Re: Film versus DSLR dynamic range

Ravncat wrote:

Though, I don't understand why people talk about an advantage of film
being how it behaves when it overexposes, unless you are going for
overexposure, it's not so hard to nail your exposure.

Really? How do you nail your exposure if your foreground is in deep shadows at IL5 and your background is bathed in light at IL15?

Fill-in flash would be the answer, but it's not always possible. HDR is another answer, but again, it's not always an option.

With B&W film, I could "nail the exposure" for the foreground and let the background wash away in the highlights. This could add to bokeh and look rather good actually. Besides, grain would avoid the "cut out" effect you get with digital.

DR is not a problem per se, the problem is with how digital behaves at the clipping point. Color shift due to channel clipping is rather ugly for instance.

RomanJohnston
RomanJohnston Forum Pro • Posts: 18,827
Nail the exposure for both areas....and do this...

http://www.nwpphotoforum.com/ubbthreads/information/php/2007_Articles/rjohnston/roman2.php

Yeah...the tutorial is using one RAW shot...but works the same for bracketed exposures.

Roman
--

The Law Of Attraction is ALWAYS working. Your only choice is whether you drive 'it'...or 'it' drives you.
-Me
http://www.pbase.com/romansphotos/

uk102 Senior Member • Posts: 1,286
Re: Thow in my digital blending technique...

DR is measurement of usable range above the acceptable noise floor. Therefore you have to decide what is acceptable and therefore it is a subjective measurement.

I find that in real world printing, it’s best to avoid using DR greater than 5 stops

arachnophilia Veteran Member • Posts: 3,362
depends on quite a lot

A number of posts in recent threads state that film has more dynamic
range than DSLR sensors. Actually, these measurements

( http://www.clarkvision.com/imagedetail/film.vs.digital.summary1.html ) clearly show the contrary, with slide film having about 5 stops dynamic range, print film about 7 and a DSLR getting to 10 stops. I would be interested in understanding on what are the statements about the low dynamic range of DSLRs based?

if we're talking BLACK AND WHITE film, and not color film like the article does, there are loads of ways to change the dynamic range of the negative -- ansel adams describes his "n+1" process in the book "the negative" for instance.

it depends on the film used, the process used, and how an image is printed. but we already know that slide film has a very small dynamic range, and the color print film ain't the greatest. the people talking about film's dynamic range are probably talking about stuff like agfapan 100 (or even 50) and not kodak gold 200.

 arachnophilia's gear list:arachnophilia's gear list
Nikon D700 Nikon D300S Nikon AF-S Nikkor 70-200mm f/2.8G ED VR Nikon AF-S Nikkor 24-70mm f/2.8G ED Nikon AF-S Nikkor 20mm f/1.8G ED +3 more
Jay Turberville Forum Pro • Posts: 12,917
Sure ..

your flame image is not a good demonstration of any
particular dynamic range capability.

...We do not have anything else to discuss.

.. you are probably right. I kind of figured that based on your original response to this thread.

http://forums.dpreview.com/forums/read.asp?forum=1021&message=24123429

My apologies for committing the sin of actually thinking you might support your assertions with evidence (like some resource to show what the luminance range of a candle flame is or spot meter values for the images you showed) or respond to the contrary evidence I supplied. I should have known better after reading that first post of yours.

-- hide signature --
Jay Turberville Forum Pro • Posts: 12,917
Re: Bit depth and dynamic range...

You are correct when you describe how many "levels" are available to
describe shades of gray in each channel for a 12-bit bitmap image.

However this does not map directly to dynamic range. It is certainly
possible to compress the tonal values of the scene that has dynamic
range more than 11 stops into this range and still have some
measurable detail in both shadows and highlights.

Sure. It is possible. But show me one manufacturer who is doing that when they record their raw data. So far every camera I know maps the sensor tonal values in a linear fashion. So you are right that it is possible. In fact, that is exactly what is done with in-camera images that are output as JPEGs and TIFFs with 8-bit depths. But since those in-camera images start with the same 12-bit linear raw data, they also can't output more than the 11 stops.

When manufacturers start outputting raw files with non-liear tone mapping, then it will take more than a simple math analysis to understand the potential for DR. But until then, it is a perfectly good way to look at how much DR digital cameras can deliver.

There is some slight non-linearity in real life in the darkest shadows which is probably due to lens flare. That probably explains why we can actually reach and even exceed a tiny bit the theoretical DR limits. You can see that slight non-linearity in this Imatest graph. It is interesting on its own merits, but I don't think it has much relevance to practical photography with most cameras. Note the otherwise linear response.

-- hide signature --
Keyboard shortcuts:
FForum MMy threads