Introduction
This paper is not an introduction to scanning and assumes you understand some of the basics (see http://www.scantips.com). The main purpose here is to be able to achieve the best possible raw scan of a traditional B&W film emulsion that can then undergo further post production in an application such as Photoshop.
I am certainly not the expert, however, I have spent a great amount of time in trying to achieve the best quality possible B&W negative scan from a Nikon 4000ED scanner that will allow high quality, large prints to be possible with dedicated B&W ink-sets (http://www.inksupply.com, http://www.inkjetmall.com). I have read, experimented and tested a lot in terms of workflow, beginning at which films, development, scanning techniques and post production techniques. I decided to write this paper to document an advanced scanning workflow which, in my opinion, has delivered the best quality possible in a traditional B&W film scan using CCD technology (for me).
As a note; quality here does not refer to corner-to-corner image sharpness achieved from film flatness – this is another separate issue. Quality here is about maintaining detail across the tonal range of an image, minimising grain issues and reducing noise as much as possible.
Additionally, this workflow may be considered by many as overkill – described as excessive-compulsive behaviour toward scanning and totally unnecessary. If you feel that way and you’re happy with your current scanning workflow – or that new digital camera combined with a batch convert to black & white Photoshop action are best results you’ve ever seen! – then this is obviously not for you. However, reading it with an open mind may also provide some additional insight to how and why certain other things happen in image processing.
What is not discussed here is additional post-scan processing to address noise that may still be present in an image, especially in a negative’s shadows. I have found that when judiciously applied, Noise Ninja from PictureCode (http://www.picturecode.com) can significantly further reduce shadow noise in high ISO B&W films that have been scanned via the follow techniques. At this point in time I have not been able to quantify the when, hows and why of applying noise reduction software to a particular image and therefore cannot provide any meaningful discussion at this point in time.
Profiling the film
Profiling film, or more accurately, profiling a specific film, is usually considered in terms of replicating the colour accuracy of the image on film as its digitised - . For B&W film, colour is obviously a non-issue and, tonality is most likely going to need additional post-processing to achieve a satisfactory final image. Therefore, the most important issue is going to be capturing as much real information as possible.
The term real information probably needs some expanded discussion as well. The histogram of a scanned B&W negative, as represented in Figure 1 would typically be considered a good scan result. The histogram exhibits no gaps, or combing, over its entire range and both the black and white ends taper down gentle indicating no clipping has occurred on either end.
Figure 1 - Generic B&W scan histogram
Modern prosumer scanners have no problem capturing the entire density range of a negative film – this implies that the scanner could actually work with even greater density. Because of this, a raw scan is likely to be somewhat centred in the histogram range – an image with average tonality attracts an average exposure from the scanner, thereby resulting in an image that represents a good capture of all the information with no clipping, albeit looking flat, sitting in the middle of the histogram and requiring both black and white point adjustments.
It is this adjustment of, especially the black point that can impose destructive results on the tonal distribution in the shadows that we want to fix.
In terms of being destructive, the primary area to focus on is in the shadows, 0% ~ 15% of the left-hand side. Figure 2 represents what tends to happen on the shadow end of a scanned image with a histogram similar to that of Figure 1 when the blackpoint is set.
Row [A] is from the scan – notice that the black-point falls short of being ‘0’ so the image will appear tonally rather flat. Row [B] represents where we would like the shadow tones to be moved to when the black-point is set.
Figure 2 – Tonal behaviour of setting the black point
However, there is more of this histogram to the right and both the white-point and the setting of the gamma position will anchor the right-hand side of the histogram forcing the shadows to be stretched out – row [C], resulting in an increased contrast over this range. In a B&W film scan, shadows frequently have detail and nothing is really pure black which implies that the mix of tones in the shadows become exaggerated and looks noisy.
Figure 3 - Full image from which test crops are taken
Figure 3 show the complete image and the crop area that are used throughout this document. The film is Fuji Neopan 1600 at an IE of 3200 in XTOL 1:1.
Figure 4a - Crop of raw scan
Figure 4b - Crop of raw scan after setting black point
Figures 4a/4b shows a real example (Fuji Neopan 1600 at 3200 in XTOL 1:1) of this; the lefthand image is the raw scan – notice it looks rather flat – as the black point is set in the right-hand crop, the stretching of the shadow tonality exaggerates both the noise and mixed tonality of the pixels in the details. Either way, this is not the preferred result for achieving a high-quality B&W image.
The initial objective now appears to be to limit the amount of tonal stretching in the shadows.
Achieving a good base-image
The results discussed here were achieved using a Nikon 4000ED scanner at 4000 dpi in 48-bit mode and 16x multi-sampling with Grain Reduction set to Light. The software involved was Ed Hamrick’s Vuescan (http://www.hamrick.com) and is necessary for the following procedure as the typical scanning software included lacks the necessary features.
The current objective is to avoid centering the histogram of the raw scan and basically set the black-point via controlling the exposure of the scanner CCD. To achieve this you need to create a Vuescan *.ini file for each film type – only a base is required as it doesn’t matter what exposure index (EI) the film is shot at. First, place a clear (completely fixed piece of the particular film into the scanner; perform a scan preview, ensure that the crop lines are inside the frame; scan preview again to ensure the auto exposure is only considering the clear base; then in the input tab select [Lock Exposure] – see below.
Vuescan Exposure Lock (Mac)
Once this is done an additional parameter box becomes visible with an exposure number in it. You can now save this *.ini file as FILM_TYPE.ini or perform additional fine tuning. Perform an actual scan on the blank frame – selecting a small crop will speed up the process – then in Photoshop evaluate the histogram. If it is similar to histogram [A] of Figure 5 showing some clipping of the blacks then this is satisfactory to save immediately. However, if it is similar to histogram [B] of Figure 5 then you may want to slowly manually increase the Locked Exposure value in Vuescan and repeat – note; do not deselect the Locked Exposure value in Vuescan while fine tuning. When your histogram is looking more like [A] in Figure 5, save the *.ini file with the appropriate film name.
What does scanning future negatives of a particular film type with the associated *.ini file loaded versus allowing the scanner to perform an automated CCD exposure per frame do for us? Figure 7 runs through the changes in the histogram of both scans. The left-hand side – histograms [A1…A3] – represents the auto exposure workflow; and the right-hand side – histograms [B1…B3] – represents the Locked Exposure workflow.
Figure 5 - Histogram of clear film-base
Figure 6 represents an actual scan of the clear film-base – in this case Ilford Delta 400 – with the un-adjusted raw-scan histogram inserted. As can be seen the results are somewhere half way between histograms [A] and [B] of Figure 5. The histogram displays a hard edge on the left side suggesting this is where the solid black position lies, however it still falls a little short of being 100% black.
To improve on this a little more; without deselecting the Exposure Lock check box in Vuescan, slightly increase the value in the RGB exposure value box; if now at 3.63 try moving it to, say 3.8 – see the screen shot of the Vuescan tab above – then recheck the histogram. The idea is to move the histogram toward the left until it just touches the 100% black point. In practice, the histogram, in Figure 6 is quite satisfactory. The amount of stretching required in the shadows from setting the black-point will be minimal and un-noticeable.
Figure 6 - Actual scan of clear film base after Exposure Locking
About Auto Exposure Workflow
As discussed previous, the histogram of the raw scan [A1] suggests an image of average tonality and, without clipping, has achieved a good capture of information in the negative. However, the effect of this has resulted in a tonally flat image falling short of achieving our black or white points.
Histogram [B] shows how both the black and white points are now set. The sliders are
moved in to establish where the pixels become pure black and pure white. In
Photoshop, under Levels, if the
Figure 7 - Automatic CCD exposure versus a calibrated locked exposure
Finally, after setting these end points, the histogram should look like that of [C]. Good enough? Not really for detailed shadows in B&W. If you look at histogram [A3], expanded in Figure 8 you can see that to achieve setting both the black and white points the raw histogram had to be stretched. Doing this increases the contrast across the image. Not necessarily a bad outcome; the image probably already looks a lot better! But the stretching usually creates much more noticeable damage in the shadows as against the highlights – discussed above.
Figure 8 - Upper 5% ~ 15% blacks are significantly stretched (increased local contrast)
The circled area in Figure 8 is the section where we see the most damage and where it would be nice to limit the amount of expanded contrast. What happens then if we follow a workflow using Locked Exposure? Figure 7 right-hand side histograms. The raw histogram [B1] is already skewed towards the black end of the tonal spectrum. In Photoshop the black point adjustment requires very little correction to set a pure black point. The white point however does require a significant adjustment [B2] with reference to the auto exposure image. This doesn’t seem to be a real issue as destructive manipulation and noise is far less observable in the highlights of a negative.
Another side effect is that with a greater stretch on the highlight side the overall image tends to display an increased gamma – it becomes brighter than desired. Therefore, a second correction is usually required here to pull the gamma down to an acceptable level – see the indicated shift of the gamma slide in histogram [B3].
Figure 9 - Pulling the gamma down - upper 5% ~ 15% blacks are minimally stretched (little to no
increased local contrast)
Once again, pulling the gamma down stretches the darker end of the histogram – that is it further increases shadow contrast and compresses the highlights. In practice, I have found that most of the increased contrast occurring on the right-hand side of the gamma marker, which arises from pulling the gamma down, is concentrated towards the mid-tones much more so than the shadows.
Even if this were to become a problem, the stretched-shadow side effect of the gamma adjustment could be mitigated via use of the Curves tool – rather that the Levels tool - through the use of points to lock movement of the deeper shadows, Figure 10.
Figure 10 - Using curves to control stretching in the shadows
Continuing from the earlier example, Figure 11 shows the same crop using an Exposure Locked profile of the film base, requiring minimal adjustments to the black point setting.
Figure 11 - Rescan using a Lock Exposure profile for Fuji Neopan 1600
Although still not perfect – remember this is probably closer to a 1.5 stop push so contrast issues have been exaggerated even before scanning – the final image demonstrates significantly better control of the noise in the shadows as against allowing the scanner to perform an auto exposure on the CCD.
The Results of a Real Example
Figure 12 is an example of Ilford HP5+ that was shot at an EI of 200 and was developed in XTOL 1:1. This image is the raw-scan directly from the Nikon 4000ED with no adjustments made at all. The inserted histogram in the upper left represents exactly what the raw-scan consists of. As expected from using a calibrated film-base the histogram is skewed towards the darker side of the tonal range.
Figure 12a - A raw scan example using a calibrated scanner exposure
The second histogram inserted in the upper right represents the same image but shows where the black and white point sliders need to be moved to before finding any pixels to set. Similar to the sample of Ilford Delta 400 in Figure 6, the black point is slightly short of 100% black from the scanner but requires little adjustment to correct. The lightest toned pixels, however, are a long way from being 100% white. In this example the white-point slider needed to be moved to a levels value of 159 to achieve a white point. This means that the lightest toned pixel from the raw scan was ~38% black:
The stretching of the histogram necessary to set the white-point in this example created no problems to the final image – not shown in this document – and imposed minimal additional stretching of the shadows.
Figure 12b - A raw scan example using a calibrated scanner exposure (cont.)
To show another example, a final, post processed image using HP5+ (IE 400 in XTOL 1:1). This image – as with all images in this paper - was originally worked as a single greyscale channel on an Apple LCD monitor calibrated to a gamma of 1.8 and a colour temperature of 6500 degrees Kelvin; they have been converted to sRGB with a gamma of 2.1 for this document.
Figure 13 - Mid-speed film example - Ilford HP5+
The primary purpose of this image is to demonstrate how clean the shadows can be and still maintain good detail. Originally, the darkest areas of this image were half way up on the left-hand side – the area inside the shed or recess depicted. A crop of this area is shown in Figure 14. It should be clear that there is minimal grain visible, noise is very low and that shadow-detail is very good.
Figure 14 - HP5+ (ASA 400, XTOL 1:1) scanned shadow detail>
The final conclusion as to the value of this technique will always remain with the individual and will most likely need practice and experimentation to come to your own conclusions but hopefully this document will help significantly in developing and improving your own hybrid analogue-digital workflows.