"Film has more dynamic range than digital." -- true?

I've never seen an Ansel Adams print, zone system and all, that couldn't have it's dynamic range improved with a little Photo Shop.

I don't know where Johnson gets his numbers; my own experience is somewhat different than what he is reporting. I use 5x4 and a one degree spot meter. All that means to this discussion is that I can accurately measure the subject brightness range (SBR) in a scene. For example, this scene measured 11 stops of SBR:

http://www.achromaticarts.com/big_image.php?path=flowers&img_num=2

I made this photograph in June at about 2:00 in the afternoon. IOW, in the brightest conditions that occur where I live. The film is 5x4 160PortraVC which I drum scanned. I can verify that the film was linear for the full range of the exposure, with no color shifts (film has gotten better in the last 25 years ;-).

I made a matching exposure with 5x4 Tri-X developed in XTOL 1:3 which is similarly linear. This was years ago, but IIRC the Dmax was around 1.6 which is very dense for me (I aim for a Zone VIII density of around 1.0 since I'm scanning).

So what I saw was no shouldering, excellent linearity, no color shifts, and 11 stops of SBR. This well exceeds Johnson's 6-7 stops for color negative film.

With digitial sensors, much depends on the size of the individual CCD sensor wells. The bigger wells translate into less noise and greater dynamic range. Phase One is claiming something on the order of 10 stops for the P45 MF digital back which has a very large sensor. Mike Collette of Better Light told me years ago that their sensor design let them capture 12 stops with their scanning backs (color, not infrared). This is possible because the backs make three separate exposures for each pixel as they scan -- the scan array is two dimensional at three pixels wide (an R, a G, and a B sensor) by however long. I don't doubt that their newest backs do even better.

So... it depends. The better digital solutions with the bigger sensors and therefore bigger sensor wells can nearly equal negative film's ability to capture a wide SBR. The smaller sensors in DSLRs seem to fall somewhere between slide film and negative film. The very small sensors in digital point-'n-shoots seem to be closer to the slide film side of dynamic range.

Anyway, that's my take on it.

My opinion... is that negative FILM does, slide film does not, and wet process print paper does not.

Kodak gold 100. I'd think there's more than 7 stops in this scene.

Dynamic range should be based on the range that can be captured on film. This is enhanced considerably by toe and shoulder effects and low and high exposures respectively. This is not to be confused with the contrast range of the output (image on film or paper). Velvia, for example, has deep blacks to nearly transparent results, but a relatively shallow capture range. Negatives appear quite flat by comparison.

Based on characteristic curves (*) published for various films, Fuji Velvia (reversal) has a useful dynamic range of 3-1/2 stops. Fuji Provia has a range of 5 stops. Representative negative films include Reala (8 stops) and NPS160 (10 stops). Tri-X has a range of 10 to 12 stops, depending on development.

Modern DSLRs have a dynamic range of about 7 stops (D2x) to about 8 stops (Canon 1dsMkII), based on test results published on DPReview.com and my own experience (D2x). These sensors have a 12 bit range of capture, expressed in the RAW mode.

Many MF digital sensors have a true 16-bit sensor. Hasselblad/Imacon claims a 12 stop dynamic range for their CFV back. Having seen the results, I'm inclined to believe it.

*The abscissa of the characteristic curves is logarithmic - divide the extrapolated range by log2 (0.301) to express the difference in f/stops.

Stephen Johnson comes to that conclusion gets from intense practice and empirical evidence.

How does the rendering of glare in a 2D RGB color environment get translated into dynamic range?

Just my own unscientific observation but I can't see dynamic range in the images posted because of this emulation of glare which doesn't necessarily translate to wider dynamic range from my understanding of the term. Using bounced lighting will change the entire dynamics of each of those images and so how can that be attributed to the dynamic range of film.

I can't understand how you can measure dynamic range using a Stouffer step wedge when light in real world shooting conditions doesn't behave that way nor does gauge measurements translate into how much useable detail can be captured.

How does knowing f-stop ranges allow you to predict how much detail will be captured in a real world scene?

With my entry level Pentax K100D DSLR all I have to do is expose darker and reduce the contrast setting and my histograms of a back lit tree taken in the shade at noon never clips to black and the hazy sky in the background doesn't either. Now the tree looks solid black on the camera's tiny LCD, but if I view zoom in on it on my calibrated display there's tons of detail captured. The lowest black setting bottoms out at 6 as read in PS's info palette.

That means my camera can map the glare of a brightly lit noon day smack in the middle of 255 levels of RGB luminance. What's the dynamic range of that?

Great stuff. I am marking this thread for further study.

I'm not arguing film vs digital here in regards to dynamic range. But I have to admit from my limited and frustrated experience using neg film on a Yashica SLR 50mm and Canon and Minolta P&S's from the mid '80's to the present, I could never get the dynamic range, definition, color accuracy and sharpness with no noise what so ever as seen in the Pentax DSLR capture at the bottom. I'm sure an expensive scanback would give even better results.

Tom,

The dynamic range has nothing to do with the amount of detail that can be captured, only the range of luminosity in the subject that produces a significant change in the output, ie. film density or image value. In digital as with film, you set the exposure level so that the areas in the scene you wish to capture lie within the available dynamic range.

Glare, which presumably washes out completely, does nothing to add to the dynamic range. If part of the image is washed out, increasing the exposure does not produce an increase in "density" or pixel level.

Noise at low exposure levels appears as a base line or fog level for both film and digital capture. In a digital image, the presence of noise is often the limiting factor in in shadow detail because it becomes so visible well above the theoretical baseline. Film is much more forgiving in this respect.

One of the striking features of MF digital capture is the "roundness" of the image - great shadow detail, highlight retention and low, low noise. The actual resolution is comparable to a better small-format DSLR.

Edward,

I equate dynamic range by what my eyes can see in any given lighting situation. I believe if I'm not mistaken the human eye has the widest dynamic range=can see the most amount of detail down into the shadows and in to the highlites.

My DSLR is showing me far more than what I've seen captured on film and even in your posted shot. What may be misunderstood here and in other similar discussions is the actual definition of dynamic range. If it has nothing to do with capturing detail then what's the point of measuring it with any capture medium.

I don't see any detail in the shadows of your posted image. In similar scenes I know my eye would see this. That is my point in regards to measuring dynamic range.

You'll also notice from my posted image that it's an extreme crop. Similar crops off 35mm negatives I've taken in the past would render the same cropped shot as simulated below.

Is your posted image the same crop level?

Martin Evening proclaims in his book "Adobe Photoshop CS2 for Photographers" page 427 that in 1994 a 4 x 5 photographer Stephen Johnson determined using a Betterlight scanning back that digital has more resolution, sharpness and tonal qualities than film. I assume just like a negative he has to enhance the original capture to get a wide range. Is this what people are calling dynamic range?

I also see people comparing smaller format digital capture to 4 x 5 film. I agree with Tim, a lot of people seem to be comparing apples to tangerines.

Tim,

The detail is there, down to the wood grain in the bridge and tree bark in the shadow of the leaves. How much do you expect to see on a monitor from a downsized sRGB JPEG? The point is that the darkest areas are not blocking like they do with a small format sensor. It is easy enough to open up the shadows, but then it would be "flat" and "digital" to the uninformed. I can't please everybody (sometimes not anybody);-)

The iris in you eye constantly adjusts according to where you are looking, based on a very small area in the retina (the fovea). What kind of dynamic range to you see when the iris can't adjust (e.g., after leaving the eye doctor)? It appears that mother nature invented HDR Merge.

What a great thread! A lot of good information here. I have harped on several threads lately about this...that negative color film can capture a greater dynamic range than consumer level digital sensors. I based my observations on side by side comparisons of film and digital based images that others have offered to show here on photo.net over the years. I have also observed, where I used to work, prints from APS-sized sensors that fellow co-workers would share. Time and time again, a 35mm film-based image will capture a greater dynamic range than an APS-sized digital sensor. But, the resolution of good APS-sized sensors are impressive!

Nice discussion, Bruce. And it is true that bigger sensors are better, just like film, although some refuse to believe it. For CMOS technology, the signal-to-noise ratio is linear with the size of the parasitic capacitance of the photodetector. And of course, dynamic range also increases with a larger capacitance because more charge can be stored. Bigger photodetectors...better SNR and higher dynamic range. Similar arguments also apply to CCD technology, and any other technology for that matter...big is good...that's why the Hassy digital backs and others like them are so impressive.

In overexposed wedding dress images C41 FILM often has a broad shoulder and digital has a hard clipping. An overexposed image can be saved in film shots and is often all 255's in and underexposed digital image. The range of stops in film can be very broad; and the overexposed region not clipped but mushed in so there is some difference in density; instead of all 255's with a typical non-raw wedding overexposed dress. A high end scan of a overexposed wedding dress allows a printer to pull detail and save a photographers ego/image where an overexposed all 255's white dress has to be photoshopped in with bits of the dress from other images.

With such as old Super-XX in sheets had this super long straight DLogE curve with little shoulder. One could overexpose and underdevelop and get a very large range of fstops maped into the negative.

With todays B&W films the amateur still loves to "push" ie overdevelop and underexpose; thus bumping the contrast and getting a lessor range of fstops mapped on to film. In still life work and higher end work folks use the 1920's and 1930's Kodak tech books methods of mapping the dynamic range of the scene to film via controlled exposure and development. A High range of fstops was shot with more exposure and less development. Adams later called this the zone system; zones were easier to understand than H&D curves or DLoge curves for the typical amateur who needed a watered down simpleton gameplan.

Pictoral FILM does NOT "clip" at overexposure; but has a gracefull "mushing/compression" of the transfer function. Sensors often go into a hard clipping; ie the brides wedding dress info is lost in a see of all 255's. Folks are throwing around terms like "dynamic range" with NO transfer function graphs; no data; just BS. The shoulder region of non pushed films can map many fstops when a sensor is full saturated; even when raw is used. With an overexpose and underdevelop scheme the DLogE curve can map many more fstops than a sensor. BUT an old trick with still life digital is to shoot two digital exposures and combine them to grab all the wanted 'stuff' in a pesky scene one cannot controll.

Even on flatbeds and film scanners the transfer function is not linear in the shadow areas or highlight areas some machines; calibration forces the "lookup table" to be changed so linearity is forced for awhile.

With an uncoated 1930's Xenar on a Retina and Superia 800; inside and outside shots seem to map all on to film even with gross exposure errors. The lens flare "fills the shadows" and thus there is this weird dreamy fake mapping of the data. Shots with an uncoated LTM lens on my Epson RD-1 have sort of this look too. An uncoated lens COMPRESSES the dynamic range of a scene; a wazoo multicoated modern lens has a transfer function closer to unity. The number of stops recorded on film or digital varies with the lens type; coating; flare conditions of the scene; useage of lens hoods.

Canon itself - in referring to their FF sensors, states in page 6 of their publication that can be found at http://www.robgalbraith.com/public_files/Canon_Full-Frame_CMOS_White_Paper.pdf - "Their gradations and dynamic range are now the equal of the best positive"
It is unfortunate that they don't provide details of testing but clearly this marketing material - one that will try to show to show their best quality, can only claim equality to slide.
BTW, they go on to state "No smaller sensor has achieved this level of performance." so all other sensors are not there yet either.

Care to show some samples demonstrating what has been put into words here so everyone can understand?

And Edward, if you expected us not to see shadow detail in the size of the image posted as a sample demonstrating dynamic range, then what are we to derive from your posted shot?

Can someone post 100% crops demonstrating film having more dynamic range than digital?

It wasn't hard for me to do it.

Fair enough, Tim. I think these areas are representative. The basic exposure was close to "sunny 16" - f/16 @ 1/60, ISO 50. The white gravel is +2, and the incident reading under the trees is -4. The black areas are much lower than -4, but do not show significant blocking.

It's hard to make a direct comparison because sensors are linear, and films is nonlinear (in its tonal response curve). What does it mean when a sensor can see so many stops of light, but the shadow end has no detail (because the highlights are hogging all the bits)?

http://www.clarkvision.com/imagedetail/index.html You might want to try reading this.

I don't understand how this discussion is so prolonged. Take your digital and film cameras out on a sunny day, shoot a scene that straddles the shade, and you will have your answer in the time it takes you to develop your color negetive film. Very few digital sensors could have taken Santiago's picture above without blowing out the exterior scene and maintained the same level of shadow detail. From a practical perspective color negative film has more dynamic range that almost any digital sensor. Dynamic range continues to be the Achille's heel of digital technology, and if it weren't the case film's demise would have been much more accelerated. The Luminous Landscape in its reviews says some MF digital backs are getting closer, and I don't know about scanning backs. Brian

OK, Edward.

I stand corrected. Film does have a wider dynamic range as seen from your sample crops.

After thinking about it a bit my shot showing the distant detail in the highlites of the apartment siding is probably attributed to distance. I doubt I could get the same detail shooting a white wedding dress close up under the same light intensity and conditons without causing the tree detail to plug up.

What does it mean when a sensor can see so many stops of light, but the shadow end has no detail (because the highlights are hogging all the bits)?

That's where 16-bit files come in handy. There are enough bits for everybody, even at the dark end of the scale. Besides, ACR (and other programs) apply a film-like curve when deciphering RAW images.

Santiago's shot has an awful lot of noise in the shadows, visible in even this size. If you lighten the shadows of digital in Photoshop, you get about a similar result. Tim's Hasselblad admittedly has an incrdible amount of detail, that only the best DSLR can provide. However, it is flawed in color representation for my humble taste. What this says to me is that photography does not depend on contrast range or detail. It's an art. Up from a certain camera level long reached by DSLR nowadays, all cameras can produce excellence.

Brian, with all due respect your first sentence is right but the rest is nonsense. Read Tim Biehn's post. I have been getting great range with Digital but some processing is required just as printing a negative normally requires manipulation.

A friend of mine believes that anyone that believes film is totally superior will still prefer the horse and buggy days. The only time I prefer film nowadays is for extended time night photography or a grain effect (usually black and white) but my opinion is changing.

That being said, whatever floats your boat as the sailors say and enjoy what you do.

I know Steve quite well. First, he's usually correct and second, never argue with him (its not at all fun). But the bottom line is, my experience fully conforms with what Steve is saying.

Also, when measuring dynamic range, there is no universally accepted method since you're supposed to start the process in the dark areas just past the point of non image forming noise. But who's to say when this starts? But, digital capture has a lot less noise in the shadows, especially when properly exposed (to the right) than film. Whenever I compare good quality drum scans of film to the same image shot digital, I'm amazed at how much grain/noise and non image forming junk I see in the film, not the digital capture.

"I don't understand how this discussion is so prolonged."

And I don't understand how folks can sit in the blinding heat on a weekend watching pretty colored cars no one can afford to own much less operate race around in a circle for an hour or so.

It's the pursuit, not the prize. But in forums like these it's a little of both and the prize being someone learns something.

That's fun to me.

Canon's white paper states what it states to prevent itself from getting badgered and possibly sued. It's the same reason why most food manufacturers provide .1 gram more food in the package than the label indicates. Same reason why pharmaceutical industries print a slightly lower toxic dose threshold on their labels.

Andrew, who is Steve?

Johnson!

Canon's white paper also throws all their none full frame DSLRs under the bus with their statement as well.
The fact of the matter is that the King of All DSLRs . . . the 1Ds MKII only has the latitude of slide. That means Canon itself, after their own testing can only reach slide film under the best marketing ploy. Marketing, by it's nature, usually means stretching the truth a little bit or minimizing the other, or both. Considering all that the best they can state is the same as slide . . .
Since the new 1Ds MKIII is not full frame it must therefore not have met the same level as their full frame either since Canon hasn't saw fit to update their White Paper.
BTW, has any other sensor manufacturer made any claims one way or the other?