Scanning and Film Latitude + Dynamic Range

[Ed_Ingold]

<p>John A:</p>

<p>Density is measured in the same way for both negatives and reversal film.</p>

<p>The relationship between exposure and density is expressed in the characteristic curve for that film. The absissa (horizontal axis) is the exposure and the ordinate (vertical axis) is the resulting density for that exposure, both in logarithmic units.</p>

<p>Film doesn't go from black to clear (white), it goes from dark to light (grey, orange, or whatever). The darkest tone is the DMax and the lightest tone is the background level (fog, film base, mask, etc.) The useable values lie somewhere inside these extremes.</p>

[sfcole]

<p>There's a lot of concern here with density, but I find the real problem in scanning color--any color films--is the ability to get the color right. The following was scanned with an Epson:</p><div></div>

[sfcole]

<p>Compare this to the same photo but scanned on a Nikon 8000 with Vuescan. Both are Velvia, yet I have been simply unable to get satisfactory color from this slide from Vuescan (I tried several different color settings and methods, to no avail). It doesn't happen consistently, just on certain photos. Consistent color balance is, to me, one of the trickiest aspects of the process.</p><div></div>

[john_a5]

<p>Scott, most of the time the difficulty is just not having a good profile for the film you are scanning. Using 3rd party software might be a bit more dicey, but do they have a profile for Velvia? If you don't have one, you might check their website or e-mail them. You shouldn't be having issues like that with slide film, at least it hasn't been my experience with 5 different scanners--my first was a flatbed and I also had one later that I used for 8x10 transparencies and it was never difficult to get a good scan from a TP--but I have never used Vuescan.</p>

[randall_pukalo]

<p>Let me clarify: Scannin slides with good results is MUCH easier than scanning negatives. A slide scans just like a digital image - it is sharp, with very little grain (and hence can be easily sharpened), great resolution, and great colors. The only difficult thing with it is if you have large areas of very dark shadows. In these areas, less expensive and/or older scanners have a harder time reading through the shadows.<br>

Negatives on the other hand scan with much more grain, lower resolution, and most challenging of all, it can be very difficult to get good, consistent colors from the negs. Unless of course, you have a machine optimized for scanning color negs, such as a minilab scanner like the Kodak f235. These read the barcode on the film, and have specific film lookup tables to get the colors right. But these are out of reach of the budgets of most consumers.</p>

[AlanKlein]

<p>John A. I love your web page photos. Your photography tells wonderful stories. Were you shooting film on the train rail shots? I noticed some have a lot of contrast which I like. You allowed some areas to go dark.</p>

<p> I find blacks add drama to photos and make them "pop". In that sense, the inability of scanners to dig through the shadows can be of value. Getting a lot of detail in shadow areas is beside the point because the main eye views are the light areas. No one really examines the shadow areas to see what they can see. I think too many people make getting shadow detail a technical challenge and forget that black adds drama and is a counterpoint to the light.</p>

[zack_zoll]

<blockquote>

<p ><a name="00YMmO"></a><a href="../photodb/user?user_id=5245443">John A</a> <a href="../member-status-icons"><img title="Frequent poster" src="../v3graphics/member-status-icons/1roll.gif" alt="" /></a>, Mar 10, 2011; 10:06 a.m.</p>

</blockquote>

 

<blockquote>

<p>The scanners also seem to like a bit thinner negative than what is optimal in the darkroom, this also gives you a little more headroom. I can shoot my film, for scanning, at the rated iso or even a bit faster, and know that I will get a good solid scan from it. Transparencies, scanned or not, have a limited dynamic range and the best results are when it is exposed properly--looks great on the light table</p>

</blockquote>

<p>I have found just the opposite John. Perhaps it's the scanner (my lab's Noritsu), but I've had much better results with overexposed images. I find that thinner negatives register too much contrast, and I have difficulty correcting it. Denser negatives almost always need to be corrected in post (usually just setting the black points), but after that I get a much broader tonal range than if I had exposed 'correctly.'</p>

<p>I usually shoot Portra 400, exposed at 100 ISO. I tend to get wide tonal range, and few or no blocked up highlights. When shooting black and white (usually Ilford FP4) I rate it at 50 if I can support it (box speed is 125), and then develop in Rodinal 1:50. I also use a temperature of 64 degrees at 15 minutes, rather than 68 degrees at 11 minutes. The combination of low power developer and long developing time gives me a dense negative, albeit one with many more midtones than a standard developing method would. I also find that my scanner prefers Rodinal to other developers; when using TMax, D76, or Perceptol, I had WAY too much contrast.</p>

<p>I cannot speak for other scanners in relation to colour film as the idea of overexposing colour is pretty new to me. But for black and white, every scanner I have used has given me the best results overexposed, and with Rodinal.</p>

 

[Craig_Cooper11664875449]

<blockquote>

<p>Craig, scanners can indeed output to raw. I use viewscan and can save in Raw and DNG</p>

</blockquote>

<p>The fact that scanner software call it RAW doesn't make it RAW. The output from a scanner is not a RAW file in the true sense, which would need to consist of non de-mosaic data</p>

[robert lee]

<blockquote>

<p>If you have a 9 or 10 stop range ... the compression to get it all will make both areas look awful ... major histogram gaps. You can't mask, burn and dodge in a scan, so putting out reasonable output for both areas is a much better solution to the problem.</p>

</blockquote>

<p>The compression happens on the film - mapping the 10 stops of scene brightness down to 5 stops of density excursion on the negative. So in this sense, it's already too late. </p>

<p>However, the point is that the scanner has no problem covering 5 stops; the entire range of tonality can be digitized to the resulting file in just a single pass. If you're outputting separately by manually overriding the detected black and white points, then the scanner driver is just being asked to do internally what arguably could have been done better in post.</p>

<p> </p>

[robert lee]

<blockquote>

<p>The output from a scanner is not a RAW file in the true sense, which would need to consist of non de-mosaic data</p>

</blockquote>

<p>Raw pixels simply implies that you're getting what the CCD detected with no further processing from the scanner driver or scanning application. Even the most basic scanner can provide a close analog by just disabling the various adjustments and outputting TIF.</p>

<p>Scanners don't demosaic because there is nothing to demosaic. Each location on the film is sampled for R, G, B (and IR if ICE is turned on.)</p>

[AlanKlein]

<blockquote>

<p>However, the point is that the scanner has no problem covering 5 stops; the entire range of tonality can be digitized to the resulting file in just a single pass. If you're outputting separately by manually overriding the detected black and white points, then the scanner driver is just being asked to do internally what arguably could have been done better in post.</p>

</blockquote>

<p>So Robert, what does programs like Vuescan that do multiple scans of the same image provide if one scan can capture the whole tonality? I've been using Epson's program on my V600 and don't seem to have any problems with one scan and do my PP in Elements afterwards.</p>

[Dave Luttmann]

<blockquote>

<p><a href="../photodb/user?user_id=408087">Craig Cooper</a> , Mar 10, 2011; 09:55 p.m.</p>

</blockquote>

 

<blockquote>

<p>Craig, scanners can indeed output to raw. I use viewscan and can save in Raw and DNG<br>

The fact that scanner software call it RAW doesn't make it RAW. The output from a scanner is not a RAW file in the true sense, which would need to consist of non de-mosaic data</p>

 

</blockquote>

<p>Craig, there is nothing to demosaic. Scanners don't have a Bayer filter like a DSLR. According to Vuescan, the raw output IS raw.</p>

 

 

[mauro_franic]

<p>"So Robert, what does programs like Vuescan that do multiple scans of the same image provide if one scan can capture the whole tonality?"</p>

<p>Average CCD noise mainly. Multipass does not provide any material value on scanners with low electronic noise like the Coolscan.</p>

[Ed_Ingold]

<p><em>So Robert, what does programs like Vuescan that do multiple scans of the same image provide if one scan can capture the whole tonality?</em></p>

<p>That's a very good question. Silverfast also offers a multi-pass scan which purports to improve dynamic range. I think this is marketing hype. Multipass scanning can improve thermal noise performance by simple averaging. Unfortunately, you need to scan in powers of two to see any noticeable benefit, and multiple scans adversely affect sharpness and increase scan times.</p>

<p>Noise improvements, if any, are found with reversal film, which stresses the ability of the scanner to "penetrate" the darkest portions. Density is never an issue with color negative film. Nikon scanners have a very low noise level, but earlier flatbed scanners may show a significant improvement.</p>

[Ed_Ingold]

<p>RAW film scans are TIFF rather than RAW files in the sense used by digital cameras. They contain all of the data produced by the scanner without significant processing, including negative inversion. Nikonscan tags the results with a NEF file extension, but they cannot be opened in Photoshop unless you change the extension to TIF. Silverfast may use a 32 bit format, which doesn't add any information to a 14 or 16 bit scan, but reduces roundoff errors in subsequent processing. </p>

<p>The main purpose of scanner RAW files is to allow use of the parent scanning software to process the results using several scenarios, without the need to physically rescan the film. This can save time if you have a variety of subject material and lighting conditions, or if you do unattended batch scans.</p>

[john_a5]

<p>Alan, thanks for the nice words. All of the train images were film, 4x5 Portra 160vc or MF with the same film. Most of the aerial work was done with 400vc and a few of the ground shots, but not too many.<br /><br />As to darks, I think the key is the balance of those areas with the rest of the image and what else can inform those areas. These are all reductions from about 200mb (8bit size) files and there is more detail in the shadows than might be seen here, but some is very subtle.<br /><br />Zack, good point as it does depend probably on the scanner and who is operating it. Doing it yourself has its benefits as you can learn the equipment's nuances and know when things change. I always overexposed my negative film for darkroom prints, about one stop, in order to boost the shadow detail. Your results, no blocked up highlights even overexposing 2 stops is exactly what I meant regarding there being a lot of range in negative material, even color negatives.<br /><br />Craig, it works like a raw file, that is all that is important to those who use it.<br /><br />Robert, I totally disagree. Negative material is designed to do the compression and it keeps the steps--maybe 12 or more-- in even increments unless you are on the horizontal part of the toe or shoulder. On the other hand, when you output a scan and try to put all of those steps into a media that is designed to only render 5 or 6 steps(even software), that is where the problem arises(it isn't designed to keep 10-12 steps in tact). What I am suggesting is that there may be two separate ranges in a negative that fit into the output media as it was designed. My experience is that if you compress all 10-12 of those zones into a media that was designed to only express 5 or 6, the image suffers--micro contrast, local contrast and fixing it really roughs up the pixels(fixed is a relative term as I haven't seen it yield something I am happy with). If you get good results doing it, that is the bottom line. I find my images suffer when I have tried doing that. That's not to say that I have never compressed output, but it depends on how much needs compressing. On the train images, I not only output a separate sky for one, but also one for a section of the cars that was otherwise blown out because of the way the sun was hitting it.</p>

[robert lee]

<blockquote>

<p>So Robert, what does programs like Vuescan that do multiple scans of the same image provide ...</p>

</blockquote>

<p>Oooh, don't get me started on Vuescan :-)</p>

<p>In any case, Edward, Mauro, et al. are right on why multiple samplings can be useful. I'll add that Vuescan didn't implement it correctly to boot (when I last looked at the program years ago at least.) Vuescan moved the CCD carriage for each pass rather than sampling in place. Scanners are imperfect, and mechanical misregistration is a given pass to pass. The resulting, multipassed scan looked less noisy in large part due to blurring.</p>

[francisco_disilvestro]

<p>Silverfast has two different functions involving multiple samplings:</p>

<p>Multi-Sapmpling: multiple scans (with same exposure) that are averaged to reduce noise. You can select up to 16 scans. </p>

<p>Multi-Exposure: Two scans (only 2) with different exposure to supposedly increase the DR of the final image. It is like an HDR. This is not needed for color negative film.</p>

<p>Anyway, I agree with the coments from previous posters about the problems related to mechanical misregistration.</p>

[robert lee]

<blockquote>

<p>... when you output a scan and try to put all of those steps into a media that is designed to only render 5 or 6 steps(even software), that is where the problem arises ... if you compress all 10-12 of those zones into a media that was designed to only express 5 or 6, the image suffers--micro contrast, local contrast and fixing it really roughs up the pixels</p>

</blockquote>

<p>The cool thing though is that the film recorded most of it and the scanner digitized all that was on it. This at least gives the opportunity in figuring out how to print it.</p>

<p>Contrast masking is a technique I find often useful for tonal compression to fit the print. The mask takes a key stroke and two seconds to make. It's certainly inexpensive enough to try.</p>

<p>The following photo was shot on Tri-X during semi-cloudy day. Still, the scene dynamic range is fairly wide. I think the picture would work better if the high lights were pushed down to draw some details in the sky, and the shadows pulled up so it's not just black blobs everywhere.</p><div></div>

[robert lee]

<p>This is the contrast mask drawn automatically from the above.</p><div></div>

[robert lee]

<p>Now, the resulting masked image. </p><div></div>

[paul_goutiere]

<blockquote>

<p>RAW film scans are TIFF rather than RAW files in the sense used by digital cameras. They contain all of the data produced by the scanner without significant processing, including negative inversion. Nikonscan tags the results with a NEF file extension, but they cannot be opened in Photoshop unless you change the extension to TIF.</p>

</blockquote>

<p>I scan with my Nikon scanner to a NEF file. I have Adobe CS3 and have downloaded "Adobe DNG Converter". This gives me a "DNG" file extension to work with in Photoshop.<br>

<br /> Photoshop will allow me to work with a DNG file as any RAW file. Indeed it appears to respond identically to RAW files from my Canon G11 or a friends Nikon D700.</p>

[AlanKlein]

<p>Robert. That was a good attempt to try to correct a gray sky. However, while you brought back some detail in the sky, the tree trunks are too light and unrealistic looking. Sometimes, you have to let a picture go or just live with some of its shortcomings.<br>

I'm wondering if we're expecting more out of our scanners than is possible? The main question I have, taking a properly exposed picture, do you or should you really have to do multiple scans and any other kind of heroic PP to "save" a picture? Isn't the problem then in the exposure?</p>

<p> Where I've noticed a difference scanning properly exposed photos with a V600 flat bed, 35mm do block up more in the shadows but 120 pictures seen to do OK whether chromes or negatives with single scans. I do have to PP to bring back the levels and colors and contrast, but nothing too heroic.</p>

[john_a5]

<p>"The main question I have, taking a properly exposed picture, do you or should you really have to do multiple scans and any other kind of heroic PP to "save" a picture?"</p>

<p>No, you shouldn't if you only photograph in "normal" conditions! The problem is that the world isn't normal. The image I referred to above in the link was what was there and it was not "normal". Yes, I could see everything I have in the print but the eyes are a marvelous device and when we get cameras to emulate their ability--and the cameras ability to see near and far and side to side clearly at the same time--you will be totally correct.</p>

<p>Yes, some use graduated neutral density filters--maybe that is what you mean, get it in camera. But the reality is that ND filters can cause issues (by the way, this image was actually the one time I used one and it still needed to be scanned twice--the range was monstrous!--you can't see the cloud with the sun in it there!). Besides, what is important is to know the tools you have available and their capabilities. That includes the film you use, meters, development lattitude, scanner and even photoshop.</p>

<p>The bottom line is always getting the information you need to get the image you want to produce. When we get that all in camera and on film in a way that requires one scan, great and if it requires 5 different scan outputs, I say great! In both cases, you were successful and, in some cases, the latter may represent the more skilled photographer as a "normal" scene doesn't require that much skill to get it all--"P" will generally do it.</p>

[robert lee]

<blockquote>

<p>... the tree trunks are too light and unrealistic looking. ... I'm wondering if we're expecting more out of our scanners than is possible?</p>

</blockquote>

<p>The tree trunks are that unrealistic looking in real life too, about the shade of newsprint. Dunno, some kind of elm maybe.</p>

<p>Anyways, the problem here isn't the scanner; it had digitized about all there was. The film itself had run out of steam. On this particular negative, the sky is about as dense as the film leader. The tonal resolution is also low because of the dynamics compression at this end of the scale. Multiple scans of the same negative won't help here. Multiple exposures, i.e., HDR type techniques, would have provided more options.</p>