Scanning BW Film and Dynamic Range

[mark_farrell]

<p>I've been shooting BW film, doing my own processing, and scanning on a flatbed scanner (Canon 9950F). Presumably shooting with a digital camera yields less dynamic range than what a properly exposed and processed BW film can do. My question is, does scanning, which is a digital, reduce or limit the original dynamic range that the film may have and if it doesn't, does the method of printing, such as inkjet introduce any limiting factors? The specs on my scanner sort of suggests it has better DR more than a good digital camera, but I've read that manufacturers tend to play games with the numbers on scanners to make it look more respectable than it really is.</p>

<p>Thanks......</p>

[jonathan_montague]

<p>not sure about the scanning part as i have mine done by a lab but for B&W printing i dont think you can get better than the Piezography ink system which has multiple shades of black.</p>

[Ed_Ingold]

<p>There are two values for "dynamic range" which concern you - range of capture (input) and range of output (density). You can obtain hard numbers for these values from the characteristic curves published by the film manufacturer. (The scales are logarithmic, so divide differences by 0.3 (= log 2) for values in stops.</p>

<p>You will see that negative film compresses the input by a factor of at least 2:1, so two stops in is one stop out in terms of density range. It is unlikely (except in filmfan fantasies) to expect more than 12 stops of capture with B&W film, which translates to 6 stops out. Most scanners have a 7-9 stop dynamic range (comparable to a typical DSLR), so you are covered with room to spare.</p>

<p>The other figure of concern is DMax - the maximum density which can be read by the scanner. Grossly overexposed B&W can have a very high DMax (probably over 6), but the useable range is generally under 4.0. Again, most modern scanners will handle a DMax of 4.2 or better.</p>

<p>Printing, chemical or digital, involves another stage of compression. A good inkjet print equals or exceeds the dynamic range you can with a chemical print on paper with a comparable finish - about 3 stops from black to blank.</p>

[rodeo_joe1]

<p>Hi Mark. I think the answer to all your questions is a resounding NO!</p>

<p>Shooting with a digital camera yields less dynamic range than B&W film. Nope! A good modern Full-frame DSLR or MF digital back gives you about 12 stops to play with if you shoot RAW, and you'd be hard pressed to exceed that with film and not have the final prints looking like grey mud. Shooting with lesser digital cameras your mileage may vary.</p>

<p>Does scanning reduce or limit the dynamic range that the (B&W) film may have? Again no! A properly exposed and processed B&W film shouldn't have a density range that exceeds 2.7, and preferrably stays below 2.5D. This is well within the capabilities of almost any modern 16 bit scanner worthy of the name. Of course, if you push a film like Tri-X and get a Dmax of 3.0 or more then it won't scan, but neither will it print easily in the darkroom.</p>

<p>If you're having trouble with your 9950, then the most likely cause is overdeveloped negatives. You might also want to take control of the exposure in the scanner interface. This particular scanner model is fond of overexposing its scans and thereby losing shadow detail. All you should need to do is choose the non-auto/advanced section of the exposure interface and pull the exposure slider down a bit.</p>

<p>Finally I'm going to stick my neck out and say "No standard inkjet printer gives Black & White output that even remotely compares in quality to a silver-gelatine print". I'll quickly qualify that. By "standard" I mean any injet sold for general use, as opposed to an expensive professional large format model that takes inks in shades of grey, and even then I think that the silver-gelatine print might be marginally better. It'll certainly have been cheaper on raw materials to produce!</p>

[rodeo_joe1]

<p>Edward, I think you'd have to try really hard to get a B&W film to give a Dmax of anything even remotely close to 6.0D. I've measured the black fully-exposed leader of a few strips of various types of 35mm film and never seen a reading much higher than 3.6D. In fact, if I'm seeing maximum densities that high, then I know that there's almost certainly something wrong with the developing procedure.</p>

<p>They're a bit outdated I know, but the curves shown in Ansel Adam's book "The Negative" are a good guide to the density range that should be expected from properly processed B&W film. Just thumbing through the appropriate appendix, I'm not seeing any "Normal" development curves where the density exceeds 2.2D.</p>

[stratvio]

<p>I think I'd differ on the dynamic range question... as far as the comparative ranges of film vs. digital, it seems the best digital sensors have ranges on the order of <9 stops. Plus, this is a linear capture where anything brighter just gets lost, in contrast to film which has the "shoulder" which gradually adjusts the response with increasing intensity. As far as the dynamic range of the scanner, I've been amazed at what I've been able to pull out of film - e.g. landscapes where I retained detail in bright clouds and hill shadows, which I may have needed to use HDR for with digital. In the worst-case example, I shot a roll on an RB67 with a lens with a stuck shutter that I didn't find until the next day - all the shots were taken at ~3-4 second exposures in full sun, and the negatives came out pure black. I put them on the scanner just to see what would happen, and low and behold - images! Noisy as all get-out, but some quite useable and interesting. For printing, I generally prefer to use wet darkroom when possible, but I've had lots of situations when the negative just wasn't suitable for wet darkroom prints, but digital gave good results (that over-exposed situation being one).<br>

http://www.photo.net/film-and-processing-forum/00Rb8S<br>

http://shutterclick.smugmug.com/gallery/6616619_YJEwK#424020444_n2LsD-O-LB<br>

http://www.digitalcamerainfo.com/content/Canon-EOS-5D-Mark-II-Digital-Camera-Review-19948/Resolution.htm</p>

[scott_ferris]

<p>Rodeo,</p>

<p>You are right, B&W film, fully exposed to light does not have a Dmax close to 6.0. Even unexposed slide film, which is much blacker, struggles to get past 4.0D, no B&W film, even test films, register whites anywhere near as dark as slide film does its darks. The negs would be so dense as to be unusable.</p>

[mark_olwick]

<p>One tip: Clyde Butcher, who many call the modern Ansel Adams for his stunning LF landscape work, says to always scan your B&W negs as positives, then flip them in Photoshop. He says that the range is much better doing it that way.</p>

[Ed_Ingold]

<p>At least one Kodak curve for Tri-X goes to DMax 3.4, but most are less, depending on development. I was guessing about 6, because you can look at an eclipse though fogged Tri-X. Don't try that with color film - dyes pass IR, which is blocked by silver.</p>

<p>Scanner software has advanced a lot in the last ten years, let alone the 14 years since Adams passed on. I don't see any advantage to scanning B&W as a positive then inverting. It does make sense to scan in 16-bit mode because 8-bit mode doesn't allow enough room for adjustments. The final B&W image can be in 8-bit mode without any significant loss.</p>

[sumo_kun]

<p>"Scanner software has advanced a lot in the last ten years, let alone the 14 years since Adams passed on. I don't see any advantage to scanning B&W as a positive then inverting" -> If this was supposed to reference Mark Olwick's post then I think you are mistaken. Mark was talking about Clyde Butcher who is still alive (?) and not Ansel ;) If I am mistaken I apologise.</p>

<p>My personal experiences with scanning B+W have been much better than trying with an ( admittedly old) DSLR. Scanning my negs on even a middle range Epson from a few years ago pulls out details I am amazed with sometimes. Printing it so it looks good is another matter though. With my DSLR in RAW, I always get blown out highlights much easier.<br>

I think that even with this huge DR, printing is going to be a challenge because you have to squash it all down again to get it into the print. I suck at PS so I will always try to print traditionally. I find it easier to get a natural result. I also find traditional prints much nicer looking than digital prints. Maybe its my crappy PS skills but even high quality digital prints from neg from a pro lab look less appealing than my own prints from the same neg. I (and just about every Japanese photographer) trust this lab *a lot* for colour work so I don't really think its their fault.</p>

<p>Anyway, to answer the original question, I think the Canon you have will be able to get all the detail from a neg providing you scan properly.</p>

[Ed_Ingold]

<p>Chuk,</p>

<p>There is no rule saying you MUST overexpose with a DSLR. As for your PS and printing skills, consider it your challenge to learn the tools of the trade.</p>

[jeff_livacich]

<p>Adams died in 1984, Edward.</p>

[ondebanks]

<blockquote>

<p>as far as the comparative ranges of film vs. digital, it seems the best digital sensors have ranges on the order of <9 stops</p>

</blockquote>

<p>There clearly are huge differences between the way that digital DR is measured in different places. The www.digitalcamerainfo.com link from Eric gives 7.4 stops for the EOS 5DII. The engineering definition (full well capacity/ readout noise) gives 14.7 stops for the same camera/sensor, or over 13.5 stops when the ADC noise is factored in (http://www.clarkvision.com/articles/digital.sensor.performance.summary/index.html). DxOmark gives it 11.9 stops.</p>

<p>How to reconcile these differences? Well, the Clarkvision values are based on real data, but nevertheless they are theoretical insofar as they make certain assumptions (ignoring noise sources other than readout). I believe that the DxOmark value is a realistic value for the <em>sensor</em>, but not for the image as seen by the photographer - DxOlabs make this very clear: "All measurements are performed on the RAW image file BEFORE demosaicing or other processing prior to final image delivery". But I'm still puzzled on how one can lose over 4 stops (11.9 -> 7.4) in the demosaicing and processing. I do wonder about that 7.4 stops value from DCI; it must depend greatly on where they (or Imatest) drew the line on shadow response. Or could it be something else? - from their description: "To test dynamic range, we take a number of photos of the Kodak Q-13 chart at a number of exposure levels" - that sounds more like a test of exposure latitude, which is not the same thing as dynamic range. But they're hardly getting that distinction wrong...?</p>

<p>Anyway, back to film scanning. You can pull out even darker details from your slide/negative if your scanner supports multi-sampling. Not sure about the Canoscan 9950F, but my Epson 4990 Pro does - not in Epson's supplied software, oddly, but in Vuescan, which can do up to 16x multi-sampling. This is not to be confused with multi-pass scanning, which tends to fail to positionally register the individual scans as accurately as it should.</p>

 

[erie_patsellis]

<p>B&W film, when properly exposed and processed, should have a dynamic range of 1.1 to 1.3 if you have spent the time to calibrate your process. Even with a Fb+fog of .5, you're still only looking at a Dmax of 1.8.Where the scanner Dmax becomes an issue is when scanning transparencies, where the Dmax can approach 3.0.<br>

While film generally has a dynamic range advantage, there are exceptions. I have a Dicomed Field Pro back, predecessor to the Betterlight, and 9+ stops of dynamic range is typical without any heroic feats, you can also save a 12 bit linear image as well and apply whatever tonal curve you desire. The downside is that it requires tremendous quantities of light to reveal those strengths. (think 2-5K of tungsten in a studio setting, or between 1-2K of HMI)<br>

Another (admittedly older) camera that has phenomenal dynamic range is the Kodak SLR/n, I measure between 11 and 13 stops consistently in a studio environment and for landscape work, it's low ISO modes (as low as ISO 6) which works by averaging multiple exposures to reduce noise, is amazing.</p>

[Ed_Ingold]

<p>Ray, the figures you give for digital cameras are grossly exaggerated. That's because they are based on the noise floor rather than the ability to resolve a useful image. The best figure I've seen for any DSLR (a Nikon D3x) was about 12 stops, using a 13 stop step wedge (q.v., DPReview) by pushing the exposure compensation feature. The value is about 9.1 stops using the same step wedge with normal image treatment. In terms of useable image range, it's hard to assign an exact number. A range of 7-9 stops is probably a reasonable figure for DSLRs. Medium format backs deliver something in the range of 10-12 stops by that criteria.</p>

<p>As a practical matter, I am more inclined to exaggerate the capabilities of film than digital, since it is less likely to trigger a mindless film v digital rant. I'm willing to grant an honest 12 stops for Tri-X in D-76, because the manufacturing data supports that number. That's as good as it gets. If people want to believe 14-18 stops for Ektar 100, let them keep their illusions. The Kodak SLR/n is arguably one of the noisiest cameras to ever slip off the assembly line. That's why nobody bought them.</p>

<p>The DMax of Fuji Velvia tops 3.8, according to the curves published by the manufacturer.</p>

[sumo_kun]

<p>Well luckily I'm not in the "trade" so I'll get by using what I like best, i.e. film ;)</p>

[swilson]

<p>One thing to keep in mind is that getting good dynamic range out of film is not automatic. Where film is mostly challenged is getting the shadow detail. I have dealt with negatives that go pretty much to clear in the shadows, when that happens nothing you do will get detail out of the shadows. To get a huge dynamic range out of a negative it is going to have to be very dark in the bright areas, depending on the scene this may not produce the best print.</p>

<p>While you are getting use to all of this I would bracket a few shoot by a lot, going up to maybe 3 stops over and see how the range between the shots compare.</p>

[rodeo_joe1]

<p>It's not really the exposure that needs to be bracketed, it's the development time. The contrast and Dmax of B&W negative film is almost wholly dependent on its development and not its exposure - as long as the exposure is adequate for the amount of shadow detail required.</p>

<p>Also a density range of 1.1 is far too low for real world use. Given a "soft" development to gamma 0.5 that gives an exposure range of around 8 stops. A more normal development to a straight line gamma of 0.66 gives us under 6 stops - well short of the 12 stops that seems to have been set as our golden target. So, to get anything near to a 12 stop SBR using normal development, the density range has to be closer to 1.8. Beside which, a density range of 1.1 would not be sufficient to expose a grade 2 printing paper to its maximum density. For example: Ilford MGiv FB needs a negative density range of about 1.5D to give its full tonal range at grade 2 filtration, and that's not allowing for any contrast loss in the enlarger.</p>

<p>DSLR Dynamic range: I can only go by the evidence of my own eyes and the tests I've done. I'm posting the results of a fairly simple test. The methodology was to generate a "step wedge" using Photoshop and a computer monitor. The steps were compared visually with known ND filters, and measured with a photometer. The maximum number of steps that the monitor would support was a 9 stop range, therefore known density patches were placed over some of the steps to extend the range. This resulted in a set of brightnesses which covered 11 stops. The camera used was a Nikon D700 fitted with the contrasty little 55mm f/3.5 Micro-Nikkor. Capture was RAW and the images processed using ACR and Photoshop. RAW capture allows in excess of one extra stop of headroom over Jpeg.</p>

<p>The attached image shows (top) the straight shot of the "stepwedge". The piece of fogged film has a measured density of 2.4D - 8 stops - and the bottom patch is a 0.3ND filter for visual reference against the monitor steps. The bottom part of the image is processed to + 3 stops over the top part, but it's the same capture and include to illuminate the higher densities and show the noise (what noise?). Count the steps. There's a clear differentiation between every step and they cover a range of 11 stops. I didn't need to use the RAW headroom, so I think it's pretty fair to add that in as well, making at least 12 stops in all. Incidentally there's some blockiness from the sRGB colour space gamma and jpeg artefacting showing up. This wasn't evident in the 16bit original. Maybe using Adobe RGB as the capture space would improve things further.</p><div></div>

[rodeo_joe1]

<p>Another more practical example, using one Kodak reflective stepwedge in full sun and another in shadow. I threw in a patch of black velvet material (centre) to extend the range, and the inset shows the same patch brightened to reveal the shadow detail.</p><div></div>

[swilson]

<p>Re bracketing the development time, clearly the longer the exposure the darker the negative, otherwise film photography would not work. And whereas it is very easy to bracket the exposure for the same scene under the same lighting conditions it is much harder to do so with development time. Adjusting the development time instead of exposure is basically pushing the film in the cases where the exposure is too low, this will most often not give as good results as a good exposure with a normal development time, if this were not the case then why not shot iso 400 film as 1600 and just develop it longer?</p>

<p>Digital has great range, as long as some care is taken to not blow the highlights. B/W film has great ranges as long as some care is taken to not loose the shadows. In both cases this is done by taking care in the exposure. </p>

[rodeo_joe1]

<p>Sorry Scott, but I must agree to differ over the importance of exposure WRT dynamic range. Of course the density of the film depends on the exposure, but its Dmax is definitely set by the development. You can expose a film as much as you like (see my fogged 35mm leader) but it will reach a hard limit set by the type of film and the development. To take this to the extreme, if you don't develop an exposed film at all and simply fix it, then you'll get no image and no density above the film base.</p>

<p>As I've already said, pushing a film is probably the worst thing you can do to extend its dynamic range. If you want to capture a wide subject brightness range and still be able to easily print or scan the film, then you MUST extend the exposure slightly and "pull" (i.e. shorten) the development time. Neither myself nor anyone else here has suggested that pushing film is a good idea.</p>

<p>"it is much harder to do so with development time" - Nobody said it was easy! Same as RAW capture is more work than shooting a crappy JPEG.</p>

[swilson]

<p>So if you need to get the exposure right then why not bracket the exposure as I first suggested. The point I am making is that to make use of the range that the film has it needs to be property exposed, and the best way to get a feel for that is to bracket so you can compare the same shot with different exposures. It does not good too have a large DR if you expose too light such that the shadows go clear on the film, bracketing development time is not going to fix under exposed film.</p>

<p>In the same why if someone was having problems blowing the highlights with using a DLSR I would suggest to them that they set the camera on auto bracket before jumping to the conclusion that digital camera blow highlights.</p>

[Ed_Ingold]

<p>The advice traditionally given by Zone Method advocates is "Expose for the shadows, develop for the highlights." The justification for this is easily seen in the attached chart. Varying the development time profoundly affects the highlight density but has little effect on the shadows.</p>

<p>It is difficult to see in this chart, which truncates data near the shoulder, but reducing the development has relatively little effect on the dynamic range. Reducing the development time does, however, make it easier to print a wide-range subject because the contrast and maximum density is lowered.</p><div></div>

[tim gray]

<p>Edward, how's the Ektar test going?</p>

[AlanKlein]

<p>I'm getting lost with the technical points. I will be shooting 6x7 MF Mamiya RB67. I'll be digitizing with an Epson V500 that scans both negative film and transparencies. I'll be shooting outdoors and landscape not people, weddings or studio. Color not B/W. I use Photoshop Elements, just the basic editing. <br>

What film would you recommend for me? Thanks. Alan</p>