Do you expose and develop film in the same way for scanning as for printing?

This question is really for those who master both activities.

In analog photography one is encouraged to expose for the shadows and develop for highlights, whereas in digital photography, it is usually best to expose for the highlights and process for the shadows. So what do you do when you shoot film with the aim to end up with a digital file?

Most of the advice out there re how to capture and develop an image on film seems to be based on the assumption that you are going to print it optically. But does the process of digitalization mean that you might want to work differently when you know you are going to be scanning (and probably further digital-processing) the negative?

There is a corollary question to that: in my first scan tests and in tutorials I saw online, bumping exposure of the digital file seemed necessary even with negatives apparently properly exposed (and I'm talking about a dramatic +1 to +2). Is that normal? If everybody gets that, where should you bump up the exposure in the scanning process of your negative, in order to keep maximal image quality? I've seen people do it in the last stage (Lightroom), which does not sound right to me (when shooting digitally, bumping exposure up in Lightroom is the best way to add noise to the shadows). Other options are at scanning time (within Vuescan, SilverFast, etc.) or at the moment you convert the negative into a positive (ColorPerfect in my case, with the "Black" slider). I've started running some tests but I'd love to hear any thoughts on that, ideally with examples of pictures if you have them so everyone is clear what we are talking about.

Many thanks for your help!

You need to adjust your development contrast to suit the enlarger type.

If instead you are going to only scan and the negative is too dense the scanner will be defeated.

You chose a scanner for density (DMAX), and may develop for lower contrast, to reduce risk of clipping high lights.

If you underexpose i.e. push you wont have nice shadows scanning or wet printing.

Film has toes and shoulders in its response curve where it compresses the sceane in shadows and high lights, digital is more linear. The conventional cubic grain (may be) softer (lower contrast) in toes and shoulders, then tabular grain film.

Using (sometimes old) single coated lenses helps all these problems even with digital sensors. They even can make the film a little more sensitive by flashing the shadows.

You can use 'compensation' developers which stretch the film dynamic range (a bit).

In analog photography one is encouraged to expose for the shadows and develop for highlights, whereas in digital photography, it is usually best to expose for the highlights and process for the shadows. So what do you do when you shoot film with the aim to end up with a digital file?

Click to expand...

The emphasized text applies to digital capture (and to slide film). Negative film (b/w or color) has considerable latitude towards overexposure, as concerns capturing the information on the film.

There is a corollary question to that: in my first scan tests and in tutorials I saw online, bumping exposure of the digital file seemed necessary even with negatives apparently properly exposed (and I'm talking about a dramatic +1 to +2). Is that normal?

Click to expand...

No! Using vuescan, in all modes except "lock colors" (deceptive naming) you can specify at which points on the white and black ends the histogram should be clipped; I use white: 0.5% black 1-2%. Similar utilities from Nikon and Minolta behave in a similar way by default. The resulting file does not need exposure "bumping"; curves are another matter, to suit your taste. Maybe you created a raw file, with all the values for denser negatives unused, and your image values sitting in the dark values.

Yes, I heard about lower contrast developing for scanning; how many writing this speak from first-hand experience, and how many just repeat? But, a normal b/w film is developed to a contrast index of ~0.6. Assuming a full range zones 0 to IX, this means (quick and dirty estimate) a max density 9*0.3*0.6 = 1.62; that is indeed approx what I measure on, say clouds or snow with a normally developed negative. What is really challenging is scanning slides, where the Dmax~3. So a Dmax~1.6 should not pose a problem.

Bottom line: Develop normally. You retain the option to produce nice darkroom prints. And investigate another scanning software, or alternate methods. The files you produce should normally have a full range of values black to white.

Attn mods: yes, this thread contains the S word; but I guess in dpug it would be unwelcome because discussing development and contrast index is verboten...

I expose as I would for either but when I develop for scan I generally err on the side of slightly less development as I find over-developed contrasty negs are the most challenging to scan for a good wide tonal range. Most often overall I'll expose about 1/3-full stop below boxed rating (depending on contrast of the scene, the more contrast than lower the rating) and then develop about 10-25% less on time (again depending on contrast of the initial scene, more contrast more pull back of time). Then when I scan I scan for a high-but but flat initial scan to capture as a wide a range of grays as possible, then dial in the contrast with levels or curves in PS.

I may be missing out on utmost quality/evenness of tones but I tend towards preserving detail as much as possible: underdevelop (less agitation than what most others seem to use), overexpose (at least 1 stop) versus throwing it away due to contrast. Always easier to add contrast later (be it in the darkroom or LightRoom), than to try and pull back detail that has been lost/hidden.

I don't know what contrast index these "good negs" are, but they end up being less work through the whole workflow than trying to max out contrast and have the scanner/printer work hard to reproduce a contrasty negative.

This is from my own experience.

bernard_L said:

Attn mods: yes, this thread contains the S word; but I guess in dpug it would be unwelcome because discussing development and contrast index is verboten...

Click to expand...

Unless things have changed greatly, conversations about development and contrast index are welcome at DPUG.ORG.

The name there may have been changed from hybridphoto.com, but that name still is better suited to the site than DPUG.

MattKing said:

Unless things have changed greatly, conversations about development and contrast index are welcome at DPUG.ORG.

The name there may have been changed from hybridphoto.com, but that name still is better suited to the site than DPUG.

Click to expand...

Agreed, in theory. In practice, I did post about this at DPUG (http://www.dpug.org/forums/f6/do-you-expose-develop-film-same-way-scanning-printing-3740/) but received no answer.

As a matter of fact, I was more interested, from the start, in answers from photographers with solid darkroom skills, because they are more used to testing various development processes with the end result (i.e. the print, whether optical or digital) in mind.

So my question is still valid, whether you prefer to answer it here or at DPUG.

Many thanks for all the useful answers so far, and for those who may still join the conversation!

Michael

Richard Sintchak (rich815) said:

... Then when I scan I scan for a high-but but flat initial scan to capture as a wide a range of grays as possible, then dial in the contrast with levels or curves in PS.

Click to expand...

Thanks for this Richard---that's exactly what I am trying to do. When you say that you aim for a "high but flat" scan, what does it mean technically? A higher "exposure" number in Vuescan?

bernard_L said:

... No! Using vuescan, in all modes except "lock colors" (deceptive naming) you can specify at which points on the white and black ends the histogram should be clipped; I use white: 0.5% black 1-2%. Similar utilities from Nikon and Minolta behave in a similar way by default. The resulting file does not need exposure "bumping"; curves are another matter, to suit your taste. Maybe you created a raw file, with all the values for denser negatives unused, and your image values sitting in the dark values. ...

Click to expand...

Thanks for your detailed answer Bernard, that's very helpful indeed. Yes, I am going for a raw file (linear Tiff, to be inverted with ColorPerfect in Photoshop). I think what happens to me is indeed that most values end up in the left-hand-side of my histogram (after conversion), and I am trying to avoid that. Or rather, I'd like to know where to make the right choice so that eventually most value stand in the right-hand-side of the histogram with minimal degradation of detail. As far as I can tell, I can do that (a) while creating the RAW file in Vuescan through the clipping points (but I don't want to lose info so I won't use that) or the (lock) exposure commands, (b) while converting the image into a positive in ColorPerfect, or (c) when working on the positive in PS or Lightroom. My question is what's the best workflow for optimal image quality in the end.

Some of this has been mentioned in earlier posts, but let me give my perspective, starting from the beginning.

With B&W negative film, the "cardinal sin" is underexposure. Where there is insufficient exposure to create silver density after development, there is nothing that can be done to recover detail in that region. Over exposure will lead to denser highlights, but unless it is extreme there will still be detail there that can often be recovered, for instance by dodging in the darkroom. Under or overdevelopment will cause lower or higher contrast, which can also be compensated for, to some degree. But, where there is no silver, there is no image, whether the negative is printed in the darkroom or, as I like to put it, using an enlarger with a computer placed between the lens and the paper.

With digital images (whether captured directly with a camera or a scanner), the biggest sin is overexposure. Once the photosite of a sensor is saturated, it is completely unresponsive to additional light, and so detail is lost as surely as in an underexposed negative. At the other extreme, if the sensor is underexposed, detail will be lost in noise, but the cutoff is not nearly as sharp as for overexposure.

With hybrid film/digital workflow, you want to avoid both sins; underexposure of the negative and overexposure in scanning, as well as underexposure during scanning. To avoid underexposing the negative, all of the standard advice applies, i.e. "expose for the shadows . . ." In principle it might be possible to overexpose or overdevelop enough to lead to noise in scanning, but my experience is that most scanners can easily handle the densities of reasonably exposed and developed negatives, which shouldn't exceed more than about 1.5.

To get the most during scanning, it is best to adjust the software controls to use the full digital range of the scanner. This can be done using the histogram display and adjusting the white and black points so that the histogram approaches, but doesn't quite reach the two ends. It's probably best to err here on the cautious side, so that nothing gets clipped at either end. If the option is available, save the image at the maximum bit depth available (often 14-16 bits).

Some people say that it is best to scan negatives as positives and invert them in the editing software. From my own experiments with Nikon scanners, scanning as a positive does seem to give a bit better dynamic range in the final file, but I'm not convinced that it makes a huge difference. I don't think that anything is gained by saving a color image.

I also don't usually try to adjust the contrast of the scan, though it may be possible to improve the final result this way. If a 14-16 bit depth is used, it shouldn't matter much. With my workflow, my scans usually look very flat when opened in Photoshop and inverted. But that is easily adjusted. Overly contrasty is more problematic.

Otherwise, I would discourage any manipulations in the scanning stage, especially sharpening. It's best to have a minimally manipulated file that you can go back to later.

As far as I understand the meaning of the term, there really isn't such a thing as a "RAW" scan file, in the sense of a raw file from a digital camera. It's just a tiff file that can be edited further. (JPEG is not so good, because of lossy compression.)

With respect to different treatment of negatives for darkroom and digital printing, my sense is that negatives that will print well in the darkroom will generally scan and print well digitally (with an important exception discussed below). The converse is likely not to be true, as it is much easier to make adjustments digitally. So, if you want the flexibility to be able to do both, it's best to optimize for the darkroom (ideally for your favorite paper).

The major difference that I have seen in comparing darkroom prints and scans of black and white negatives is that scanning can greatly increase the appearance of grain. There are a number of discussions of this on the web, where it usually described as "grain aliasing", suggesting that it has to do with overlapping patterns in the grain and the scanner sensor. I'm pretty sure that this is NOT the correct explanation, but the phenomenon is definitely real. What I think happens is that the places *between* the grains transmit so much light that individual sensor sites are saturated, sharply accentuating the grain pattern.

In any case, the grain effect is real, and needs to be dealt with, most easily by the choice of developer. For a few years, I have been using Pyrocat HD. What others have suggested, and I think is probably right, is that the catechol stain increases the density between the grains and reduces the effect I suggested above. I'm still a bit mystified by why staining developers don't reduce acutance (sharpness, loosely), but the results with Pyrocat are excellent in my experience. I have also had good results with Ilford DD-X and Crawley's FX-37.

I hope that is helpful (and that I won't be banned from APUG).

David

Wow David, thanks so much for the very very clear summary and the advice!

I do hope no one gets banned because of me. In fact I've long looked all over the internet for such info spanning capture, development and scanning processes, but could not find it. As I wrote earlier, my guess is that darkroom masters are the ones who likely have the right expertise for this issue.

my experience is that most scanners can easily handle the densities of reasonably exposed and developed negatives, which shouldn't exceed more than about 1.5 (...) my sense is that negatives that will print well in the darkroom will generally scan and print well digitally

Click to expand...

Concurs with my statement (page 1, post #3). Also agree with other statements by dpgoldenberg.

There are a number of discussions of this on the web, where it usually described as "grain aliasing", suggesting that it has to do with overlapping patterns in the grain and the scanner sensor. I'm pretty sure that this is NOT the correct explanation, but the phenomenon is definitely real.

Click to expand...

Indeed the phenomenon is real, and well known in the domain of signal or image processing. Full explanation would be too technical in the present context. If interested, see http://en.wikipedia.org/wiki/Aliasing, in particular the section Folding. In a nutshell, noise beyond the Nyquist frequency is folded (added) onto "true" low frequencies, if it has not been filtered (suppressed) before sampling. Same issue with D***L cameras and the anti-aliasing filter.
Bottom line: Dedicated film scanner: scan at the maximum optical (not interpolated) resolution; reduce resolution after (if needed). Flatbed scanner: scan at 3200, downsample to 1600 (true resolution of most flatbeds).

Thanks a lot for this Bernard! Do you also recommend a staining developer to improve scanning, or in your opinion it's not going to make a significant difference?

Sorry, no experience with staining developers. I would rather go by dpgoldenberg's advice: "it's best to optimize for the darkroom". My view re: grain, scanning, etc... is that for b/w scanning is just a convenient (but not equivalent) substitute to contact sheets. Also a way to share images. Also, grain is an integral part of the b/w image in a print; has been said (e.g. Thornton) to enhance the impression of sharpness (not to deny that aliasing can make it worse in a s**n).

Bernard,

Actually, this is the part of the conventional wisdom that I don't think is right:

bernard_L said:

In a nutshell, noise beyond the Nyquist frequency is folded (added) onto "true" low frequencies, if it has not been filtered (suppressed) before sampling. Same issue with D***L cameras and the anti-aliasing filter.

Click to expand...

Moire patterns are generated by overlapping two regular patterns, so that they show up in digital images of regular patters of the proper frequency. Noise, by definition, has no defined frequency, so I don't see how it can generate a moire pattern. Alias peaks arise in discrete Fourier transforms (DFT) when there is a signal with a frequency greater than the Nyquest frequency. If the DFT is back transformed into the spatial (or time) domain, a new signal with the lower frequency will appear. But a DFT of a noisy signal just generates noise in the frequency domain, and noise again when it is back transformed. In the case of an optical digital sensor, the discrete sizes of the sensor elements essentially bins the captured photons and should actually reduce the noise by averaging.

Anyway, that's what I think. If you can point me to a source that explains how noise can generate "aliases", I would appreciate it.

David

Thanks both Bernard and David--I hope this discussion will provide good info for many photographers looking for it.

Anybody has scans of tests shots developed with Pyrocad HD vs. another developer? I'll try to do tests myself and upload them but it will take a while.

David,

I'll respond by PM, since this is maybe becoming too technical to be of general interest. Not right now (got to leave for work).

Bernard

I use 400 ISO B&W film for scanning and wet printing. If I push the film, I use Diafine. If not, I use D-23 1:4 20 min at 68 degrees F. That gives me a neg that prints between grades 2.8 and 3.5 on a Focomat V35 with multigrade head and scans very well on a Nikon negative scanner. The shadows are well developed and the highlights are not blown. Grain is uniform and small. I haven't done a lot [make that no] scientific comparisons. I have tried other developers and dilutions over the years, including replenished X-tol when I was doing a lot more film every week. The process I've described works for me.