Once upon a time, I posted a set of tutorials regarding my method for digitizing film photographs. For a variety of reasons, I took the tutorial down. However, I have noticed that there is still a huge amount of confusion regarding these methods, and a lot of disbelief that scanning film with a digital camera could yield results that are comparable to what can be achieved with a high-end desktop film scanner.
So, with that in mind, I have decided to resurrect my tutorials. Now, instead of simply re-posting the old tutorials, I have completely re-written them, with all of the updated methods I have accumulated over the past 2 years of scanning film in this way.
But first, a little digression (skip ahead for the tutorial)...
To all of the naysayers who claim that this method doesn't work or cannot work in theory, I invite you to look at the results that I (and others) have achieved. Yes, virtually all digital cameras use a Bayer array instead of individual red, green, and blue-sensitive sensors, like a scanner does; yes, most digital cameras do not record 16-bits/channel, like a scanner does. None of this matters. The results, I feel, speak for themselves. Who cares if something works in practice but not in theory?
Another misunderstanding of DSLR scanning is that the dynamic range captured on film cannot be accurately photographed by a digital camera, even one with "only" 14 stops of DR, such as the Nikon D800/D800E/D810, and others. This results from a complete misunderstanding of how film - particularly negative film - actually works. Did you ever notice how low contrast your negatives look - especially your colour negatives? This low contrast is real; during development, the dynamic range of the scene you photographed (which could easily be >14 stops on a bright sunny day) was compressed into a very small density range on the film; perhaps 9 stops (a density range of 3.0), which is more than adequately covered by any modern DSLR or mirrorless camera. Those densities are then mapped to the appropriate 48-bit RGB value (or 16-bit luminance value, in the case of B&W scans) during post-processing. And since an 8-bit JPEG is the most common output format, you will end up re-compressing those tones into the final image.
Anyway, the point is that a modern DSLR is more than adequate to digitize any colour or B&W negative, with respect to colour and dynamic range fidelity.
Scanning slide film is another matter. Slide film is fundamentally different than negative film; it is not simply a positive version of negative film. Slide film has a "baked in" contrast curve and the slide itself is the final product, whereas, as we've seen, the final contrast curve of negative film is established by mapping tones during post-processing.
Slide film is notorious for having a small dynamic range - the range of tones from pure black to pure white - that can be captured in a frame of slide film is very small; perhaps 5-6 stops. Put another way, the film compresses the dynamic range of the scene (which, as we've seen, can be well over 14 stops) into 5-6 stops. The photographer's job is to ensure that those 5-6 stops overlap with the important elements in the scene, and the tones outside the range clip to either white or (preferably) black.
Because the scene dynamic range is compressed into a very large density range, scanning slide film is a much more challenging affair than scanning negative film. This is because with negative film, the scanner (or digital camera, as we will see) does not have to "dig" into very deep shadows on the film (highlights in the inverted positive) to extract the detail, while on slide film, the shadows are much, much denser. The result of this is that lesser-grade scanners (like most flatbeds and even some dedicated film scanners) simply do not have the ability to extract any meaningful data from the deep shadows; this is particularly important in very dense film stocks, such as Velvia 50, which may have a maximum density of 4.0 (12 stops). For a DSLR like the D800, where dynamic range exceeds 14 stops, this is not a problem as long as the exposure is set correctly. I cannot stress this enough: scanning slide film requires patience and a lot of up-front work (on any system, scanner or DSLR) to ensure the maximum amount of detail is extracted from the source film.
Phew. That was quite the digression. Anyway, on to the tutorials.
1. The Tools You Will Need
The basic tools for my DSLR scanning method are as follows:
- a DSLR or mirrorless digital camera, capable of live view and Raw capture (and preferably capable of tethered capture, as well). Either APS-C or full frame are fine. I use a Nikon D800.
- a macro lens, preferably one capable of 1:1 reproduction. Any focal length is fine; longer focal lengths require more working distance between camera and film. An autofocus lens is also useful, but not essential. I use a Tokina 100 f/2.8 Macro.
- A sturdy copy stand. I cannot overstate how important this is. You need to keep the camera absolutely still and absolutely level (in both X and Y axes) to the film; without this, you'll get blurry images, either because the camera wasn't still or because focus was uneven across the film. A good copy stand, such as the Kaiser RS-2 (which I use) will do this. A poor copy stand will flex too much, and you'll be fighting to keep things level. Do not skimp here. If you're allocating a budget for this, spend more on this and less on your lens or camera. Seriously. One potential substitution here is to use a tripod with an inverted centre column (many Manfrotto tripods can do this). This will keep things still, but the 'level' issue might still be a problem.
- A bubble level. See above for why you need this. Get one of those small ruler-type levels that carpenters and plumbers use. The hotshoe mounted ones are ineffective, because there is too much play in the shoe. Ensure that the camera is level with the film holder.
- A light source. I highly recommend the Artograph Lightpad, which supplies bright, even, daylight-balanced light. An iPad will work, but you'll need to diffuse the light somehow or get the film elevated well away from the screen, or else the pixels will be seen in the scans.
- Film holders. You can use the holders from a scanner OR (and I recommend this) find yourself some film carriers from a disused enlarger. These are widely available on eBay. Get one carrier for each size of film you will be scanning; I have 35mm, 6x6, 6x7, and 4x5. These will hold the film much flatter than virtually any scanner film holder, except *perhaps* the magnetic Imacon holders.
- A remote shutter release OR tethering software with remote triggering. You need to release the shutter without touching the camera. I use Smart Shooter, which allows for fully-remote operation of the D800 (and many other cameras), including aperture, shutter speed, ISO, shutter release, and tethered live view.
- Image stitching software. This is purely optional. The idea is that for film formats larger than 24x36 35mm, taking multiple macro images and stitching them together provides additional resolution. For example, a 6x7 negative has ~4.9x the area of a 24x36mm negative, and could therefore be covered by 5 individual images, which could then be stitched together to provide a very high resolution final image. With only one exception, I do this for only very few negatives, and only when I want to print very large. The one exception? Large format 4x5. The reason I stitch with 4x5 is that (a) I cant get sufficient distance above the film to fill the D800's sensor with the full image and (b) when I'm shooting 4x5, I'm doing it because I want maximal detail. I use the built-in stitching function of Lightroom CC, which has never let me down. The advantage of using Lightroom is that the product of your stitching is a DNG file (i.e. a raw file). The other program I have used is AutoPano Giga, which is very good and offers advanced stitching capabilities, should you need them. In practice, however, Lightroom is generally sufficient (and you probably already have it).
- Photoshop or similar software capable of editing 16-bit TIFF files (if you're using Lightroom, I *highly* recommend Photoshop for its tight integration with Lightroom). Sadly, Lightroom is not an appropriate tool for inverting and colour correcting negatives (it is sufficient for positives, though), so you'll need Photoshop or similar.
To be clear, you can accomplish much of what I'm about to discuss with simpler tools. But - and I want to make this perfectly clear - your results will suffer. You may be perfectly happy with your results either way, and that's fine, but the method below is the surest way I know of to achieve maximal quality from a DSLR scanning rig (and yes, I know about the $50,000+ Phase One repro system).
So, let's get on with it.
2. Scanning Colour Negative Film with a DSLR
First off, let's tackle colour negatives. I suspect that for most people shooting film, this will be the section of greatest interest.
The first thing to recognize is that scanning colour negative film is not difficult. At all. The reasons why are discussed above (you didn't skip the digression, did you?).
What can be difficult is inverting and colour correcting your scans. Fortunately, I have a workflow that is highly effective for this.
I'm going to assume that you've got everything set up something like what is pictured in Figure 1.
I'm also going to assume you have your camera set up in live-view or (preferably) tethered live-view mode, that you know how a histogram works, and that you can use a Raw editor such as Lightroom.
- Set your camera to f/8, ISO to the lowest native setting (ISO 100 on the D800, for example; ISO 200 on Fuji X-series cameras). Set metering to manual, capture quality to Raw, WB to auto.
- Position your film in the holder such that a piece of unexposed film is visible; this can be from the leader, the space between frames, etc, but absolutely, positively must be unexposed (Figure 2).
- Set shutter speed such that your histogram is pushed to the right, but is not clipping. With negative film, the red channel will clip first, then the green, then the blue; if your luminance histogram is not clipping, then you can be sure that the red channel is not clipping, and therefore that neither green nor blue are clipping, either. However, using the luminance histogram alone is not ideal, because you may actually be far away from clipping the sensor. A better way is either to make test exposures around the selected shutter speed and then test the raw files for red channel clipping, or to use an RGB histogram (such as the one in Smart Shooter). In this way, you can find the exposure (to within 1/3 stop) that maximizes RGB channel exposure without clipping. This reduces (digital) noise and allows for more effective pixel pushing in post.
- Once you've discovered this optimal shutter speed, write it down. You will image every single frame of this film type. You need to repeat step 3 for every film type you shoot, because the density of each film will be slightly different.
- Also, once you've discovered the optimal shutter speed for a given film, save the corresponding Raw file and create a Lightroom White Balance preset, using the WB eyedropper to select a region of the unexposed film. Name the preset after the specific film type (e.g. 'Portra 400 WB', 'Fuji 400H WB', and so on). Repeat for every film stock you use (Figure 3).
- Using the optimal exposure settings you've now established, image your film (NOTE: make sure you give your film a good once-over with a bulb blower to remove large dust particles). If you have a capable lens, autofocus on an area of the image with suitable detail (live-view AF generally requires suitable contrast, so don't try to AF on an area of pure blue sky, for example). Repeat for each image on the roll. For a roll of 36 24x36 images, this should take about 10 min. Each image should be saved as a Raw file, and then imported into your Raw editor of choice.
- Open the Raw images in Lightroom. Crop as needed and apply the appropriate WB preset. Also apply the 'Neutral' profile for your particular camera (bottom of the 'Develop' module; Figure 4).
- Open an image in Photoshop (CMD-E on a Mac; CTL-E on Windows) and run the "Colour Neg Inversion" action within this action set (Figure 5). Briefly, this action does the following
Inverts the image to a positive using an 'Invert' layer
Normalizes the red, green, and blue channels using a 'Curves' layer.
Creates a second 'Curves' layer to allow for fine tuning of RGB contrast. This is sometimes necessary if the source image is particularly dark or (more commonly) bright.
- Switch to the 'Background' layer and remove any remaining dust spots with the 'Content-Aware Fill' tool (Figure 6).
- Tweak the colours, contrast, etc to your delight in Lightroom. On occasion, you may find the auto colours are not as you desire. This is not a fault of the method; it's simply that over or underexposure, old film, poor development, or any combination of these can affect the balance between red, green, and blue channels. Fortunately, we can easily correct this.
And that's it. Really. Once you've got this workflow down, the whole thing is very, very quick. I would estimate that it takes me roughly 10 min to 'scan' a roll of 6x7 negatives (10 images, total), 5 min to prep the images in Lightroom (WB, crop, etc), and about 5 min to do the colour conversions and remove dust in Photoshop. Total time: 20 min. 25, tops. And the end results? See for yourselves.
3. Scanning B&W Film With a DSLR
Scanning B&W negatives follows virtually the same procedure as scanning colour negatives, the only difference being that for B&W, you use the "B&W Neg Inversion" action within the linked action set (see above), which, for obvious reasons, does not include any colour adjustment layers (Figure 7). Adjust overall and local contrast in Lightroom, and enjoy.
4. Scanning Reversal Film With a DSLR
And now we come to the big one: reversal film. Unfortunately, slide film is becoming something of a rarity these days; I say "unfortunately", because slide film offers a fundamentally different take on film photography vs negative film, and the results, in my opinion, are truly remarkable. Anyway, if you are shooting slide film, or if you just have old slides you'd like to digitize, here's how to do it.
- As with negative film, you'll need to establish the optimal exposure. Again, set f/8, lowest native ISO, etc, etc.
- Setting optimal exposure with slide film is harder than with negative film because of the steep density curve involved. The best way of doing this is to get yourself an IT8.7 target for your specific film type(s); I highly recommend the targets from Wolf Faust (Figure 8).
- Photograph the target that matches your film type (I shoot mainly Provia 100F, so I choose the Provia/Astia/Sensia target), using settings that bracket your best 'guesstimate' of the optimal shutter speed. I'd shoot +1 stop on either side, in 1/3 stop increments, giving 7 total images. of the target.
- Open these images in your Raw editor and white balance the image against the grey stripe below the "11" patch (NOTE: any grey stripe will do, but this one is actually 'middle' grey; Figure 9). IMPORTANT: Note this white balance value. You'll need it shortly.
- In the Lightroom develop module, move your cursor over the patches of black on the far right of the target shot at your 'guesstimate'. Lightroom will display the R, G, and B percentages immediately below the histogram. The black and very dark grey patches (patches 20-22 + black) should be distinguishable in their RGB values from one another AND R should (roughly) equal G and should (roughly) equal B (since this is a grey patch) (Figure 10).
- Repeat this on the white patches (1-3 + white), and ensure that these patches are distinguishable from one another.
- Once you've determined that you can distinguish the blackest shades from each other and the whitest shades from each other, repeat this exercise on the remaining images you took in step 3. If any of the images show clipping in the highlights of the pure white patch, that exposure is too high. You're looking for the brightest exposure that does not clip the test chart.
- Once you've found this exposure, you will be using it for every image shot on this film type. This will ensure continuity and consistency of colours and contrast throughout the roll (and between rolls).
(NOTE: The intended use of the IT8.7 charts are to generate calibration profiles for each film type. I have tried this with DSLR scanning, and it makes very little difference AND is a headache, because it requires converting your scans to linear TIFF format using MakeTIFF. No thank you.)
- Image your reversal film at the established settings, setting a custom white balance in your camera, based on the WB you established in step 4. Since you've used a target to determine the optimal exposure, you will be able to extract the maximum amount of shadow and highlight detail that your particular camera allows; any loss of detail at this point is likely due to over/underexposure of the film itself.
- Open the scans in Photoshop (ideally via Lightroom, as above) and apply dust removal, as necessary, using the 'Content-Aware Fill' tool.
- Create a curves layer (included in the 'Positive Film Processing' action) and apply a very slight "gamma-correction" curve, as shown in Figure 11. This will help to open up the shadows very slightly
- For most slides, this is the end of the process. From here, you can edit as necessary (local adjustments, sharpening, etc). For some slides, however, you may need to apply a bit of editing to get the scan to match the slide exactly. This usually affects particularly dense and/or underexposed slides. No one said shooting reversal film was easy, now did they?
And there you have it. The results? See below.
5. Advanced Techniques
One of the major advantages of DSLR scanning is the ability to stitch, to create very high resolution final images. For example, the image below is a 9 image stitch of a 4x5 Portra 160 negative, yielding a 74 megapixel final image (Figure 12A).
And here's a 100% crop of the above image (Figure 12B).
The fine detail here is absolutely remarkable; the scan has resolved detail in the sails, which are a tiny fraction of the overall image. I have a large print of this over my desk, and the clarity is unreal. Frankly, I think I could push this even farther, either by stitching more images (say, 12 total, with less overlap between frames) and/or by using an even higher resolution camera such as the Canon 5DSR (50MP).
Stitching is remarkably easy, especially now that Lightroom has a built-in stitching function.
So there you have it; my definitive guide to scanning film with a digital camera.
Let there be no doubt: scanning film this way produces truly excellent digital files, with the not-insignificant advantage of allowing editing on a Raw file, rather than on a TIFF or JPEG.
You'll note that I haven't said anything about cost. While this method isn't expensive, it's not as cheap as, say, a flatbed scanner. On the other hand, DSLR scanning gives you the potential to produce scans that rival the very best dedicated film scanners; my informal testing suggests resolution that is at least on par with a Nikon 9000ED (which regularly sells for north of $3000 USD on eBay) and very close to an $11,000 Imacon X1 (with stitching, this method actually out-resolves an X1). In terms of dynamic range, this method is absolutely competitive with the X1 (which is itself noticeably superior to the Nikon).
So, no, this method isn't as cheap as many dedicated film scanners, but the results are outstanding, and the price is reduced drastically if you already have a digital camera (and, of course, you can actually use the digital camera for other things, should you choose to).
So my conclusion is this: DSLR scanning offers *by far* the best price:performance ratio available to film photographers in 2016.