When using a digital camera, processing your images is a fairly well documented process these days, and at its simplest can be just a few short steps between the camera and sharing your photo on Flickr or wherever. If you’re using film however, there is an extra step required – getting your image into a digital form by scanning. There are many ways to scan film, of course, and the method you choose will depend on the type of results you want, the equipment or services you have access to, and your budget and personal preferences. So with all that in mind I’m going to describe here how I scan film using a flatbed scanner, and hopefully that will be useful to some others out there.
As I posted shortly after I bought it, my scanner is an Epson Perfection V500 Photo, a flatbed scanner with a film unit. The film unit is a light source built into the lid of the scanner to shine through the film when scanning. This is essential when scanning film since a normal flatbed measures light reflected from whatever is being scanned – this is fine for things that are meant to reflect light, but film is designed to transmit light so gives poor results when scanned with reflected light.
Scanning obviously requires software, and I make do with driving the Epson TWAIN software directly in Photoshop CS3. Before I had Photoshop I did the same thing in Paint Shop Pro, although that was more limited in the formats it could accept. In Photoshop you go to File > Import… > Epson Perfection V500 … which then loads the Epson driver software. The first thing I recommend is to switch to “Professional Mode” and do everything manually. The automatic mode can save a lot of time, particularly when scanning 35mm film strips, but needs well-defined frames to correctly identify them, and then uses quite conservative settings. All fine for seaside family holiday snapshots, perhaps, but for most situations a bit more control is required.
I’m going to show how I typically scan a colour negative, but the process is essentially the same for colour positive film and little different for black and white negatives. The settings I use are shown here on the right. The source settings are “Film” to specify that we’re using the film unit, not the normal flatbed and “Colour negative film” for film type. This is about the only option I use that lets the Epson software do anything automatically as it inverts the image colours and also takes care of removing the orange mask. This could be done in Photoshop after scanning as a positive image, but letting the scanner software do it means we get a positive preview image, which is more useful to me.
For colour positive film – slide or transparency film – the setting to use is obviously “Colour positive film”. Less obviously, this is also what I use to scan black and white films, since sometimes that gives them a nice tone because of the slight stain to the film base from the anti-halation dyes. If not it’s easy to get rid of later.
For the output settings I always set the image type to “48-bit Color” – that’s 16 bits per channel. That makes for large files, but gives the most colour information allowing for better control over the tones and gradients of the final image. For black and white I’ll either keep the image in colour if I like the tone or, more usually, convert it to a 16-bit grayscale image based on only the green channel in Photoshop later. I started doing this after reading this article which makes a convincing case that the green channel is sharpest of the three.
I scan at 2400 pixels-per-inch, because that’s a good starting size for me and suits 6×6, which I shoot most often these days. It provides plenty of resolution should I want to print – printed at 300dpi a 6×6 image scanned at 2400ppi will be about 18 inches square. It also gives plenty of pixels for the rare occasions I feel like cropping.
All the other options I leave alone for now. The target size is irrelevant really since all this can be manipulated later. The various enhancement options – unsharp mask, dust removal, etc – I would rather do in Photoshop on as high a quality scan as I can get, rather than degrade the scan before I have a chance to do anything myself. So with the settings chosen it’s time to place a strip of film in the correct holder on the scanner glass and hit the preview button. One very important thing is to make sure the cutout area at the top of the film holder is not obstructed in any way. The scanner uses this area to calibrate the light source, so if it’s even partially obstructed, at best the colours and levels will be off, at worst the scanner will lock up and you’ll have to turn it off and on again.
The preview window generally looks like this:
I have used the marquee tool in the preview window to select a single frame to scan. Note that I have selected a good margin around the frame rather than trying to get just the actual image frame as tight as possible. There are several different reasons for this; the first of which is that the marquee tool is not very accurate and so I like a margin of error, to be sure I don’t accidentally cut anything out. Secondly, a personal reason, I like to crop the final images square while the actual exposed frame from my Bronica is slightly taller than it is wide and can also be slightly rotated (due to a dodgy film back insert) so I need to scan as large as possible to make sure I can crop the square I want. Lastly scanning the unexposed film rebate helps with getting the colour balance of the scan right, as I’ll talk about later.
Examining this preview, you can see that using the default settings the preview image looks poor, even when I’ve selected just a single frame with a marquee. It’s very flat, and seems to be dominated by a green/blue colour cast. The reason for this can be seen by looking at the image histogram.
You can see from the histogram details that the image itself in this case has a range of levels that is fairly concentrated in the centre of the overall range, and that the default settings set the black and white points far to either side. In addition the output levels – i.e. those the scanned image will have – don’t use the full available range either, so the whitest pixels in the output end up light grey and the blackest end up not quite black.
I always reset the output so that the black point is 0 and the white point is the maximum 255, then I adjust the input black and white points so that they more closely match the preview histogram. I do a first pass with the combined adjustments, and then check each channel separately to make sure nothing is clipped. You can check by using the Show Output button which gives you a preview of the final histogram – if there’s a big black vertical line at either end of the histogram you may have done too much. Again, because the both the histogram displayed and the levels controls are a little approximate I like to leave a margin of error and then tidy up once the image is in Photoshop. Here’s the histogram after I made my adjustments:
It’s not easy to show an example of the individual channel adjustments without this post drowning in screen captures, but you can see from the tone curve section of the window that there are now three lines visible – one each for R, G and B – due to the slightly different levels adjustments I made for each colour.
Note how much closer the black and white points are to the actual values in the histogram while still leaving a little margin. Also note I left the center slider (gamma) alone. I don’t always do that sometimes I’ll use it for a particularly dark or bright image, but mostly I’ll leave it as is, and adjust in Photoshop. So now let’s look at the preview of the image with these settings.
Comparing with the original preview this is obviously much better. The rebate appears black, the image is brighter and has more contrast and the colours appear much more natural. Now is the time to hit the big Scan button and get the image into Photoshop. This usually takes a minute or two for a 6×6 negative at 2400ppi, simply because of the amount of data – 100MB or more. Pretty much the first thing I do once the scan is in Photoshop is save it. I just save as a Photoshop format for simplicity, but you could save as a TIFF or DNG – anything that supports 16-bit colour. Then I’ll spend some time getting rid of all the dust using my graphics tablet and the clone tool, and save it again.
A quick word about dust – it’s inevitable. No matter how clean you think you’ve got everything a bit of dust will be in the scan. You can minimise it by keeping the scanner glass clean and brushing and bowing the film, but you’ll never get it all. That tiny bit of lint that left a 25 pixel long mark on the scan? It’s only 0.25mm long and pretty much invisible to the naked eye. Scanners these days, including the V500 have built in software for dust removal, but I don’t use it because firstly it doesn’t work on black and white film, due to the silver content, and secondly it has a tendency to soften the scanned image and either false identify as dust or introduce artifacts around fine detail in the photograph. Manual dust removal with a graphics tablet is by far the best method, if a little tedious.
Skipping the dust removal for now, the image looks like this in Photoshop:
The first thing to note is that there is a strip of white a few pixels wide down the right hand edge which is due to the imprecision in the marquee tool in the scan preview. You could possibly improve this by zooming the preview, but when scanning a whole roll I find it more convenient to keep the preview zoomed out, since I’ll be cropping anyway.
Examining the image itself, it looks ok, but now we’re seeing it a little closer it still seems a little flat, and with a slight blue-green colour cast. We can see the reason for this by zooming to actual pixels and taking a look at one of the corners.
In the corner you can clearly see the unexposed rebate around the frame. This is unexposed emulsion and so it ought to be black, but instead appears as a grainy mix of red, green and blue because of the combination of base fog and the orange mask. In the channel histograms you can clearly see a corresponding tall peak at the left-hand end of each one (I’ve roughly marked them). The peak in the blue channel is furthest to the right which is why we see a slightly blue colour cast – this signal is distributed across the whole image, but we can’t really see it for what it is except in the darker areas, and particularly the unexposed rebate.
The same thing happens in B&W negatives, and also in positives, although in the latter case it’s noise from the scanner itself in the dark areas. This is why I scan the rebate as well as the image – so that I can properly correct for it. I do this by simply adjusting the black point in each channel separately in the Levels tool. I just drag the black point to the right until it roughly under that left-most peak. Sometimes the peak isn’t very distinct in one or other channel but you can judge it by staying at 100% zoom and watching the colour of all that grainy noise. As each colour is adjusted the rebate turns to a more solid black, as you can see in the following sequence.
Before hitting OK, I usually zoom back out and toggle Preview on and off a few times to make sure it looks ok and make tweaks if necessary. Once that’s done we’re left with the adjusted image.
The image now has much richer shadows and more natural colour. At this point I usually save it again as a new file, so that I always have the option to start over if I’ve over done it. Then I crop the image to square, throwing away the rebate and maybe adjust the curves a little, although more and more often I find I don’t need to with film images.
So that’s basically it – my film scanning workflow up to the point where the scanning process turns into post-processing. Hopefully it makes sense and some of you will find it useful.