Big Bubbles (no troubles)

What sucks, who sucks and you suck

Photo Scanning Workflow

When I first got my film scanner, it took me several weeks and many attempts to produce a satisfactory scan. I’m sure everyone goes through the same thing, but there seems to be a shortage of simple how-to guides or example workflows for beginners, exacerbated by the fact that everyone uses different software. So here’s my current workflow; not perfect and certainly not professional, but it will get you started.

(NB. I use VueScan for scanning and the GIMP for editing under Linux. The first is reportedly better than any other scanning software, while the second is so limited compared to most commercial packages that any other software should be capable of the same functions.)

  1. Blow the dust off your slide or neg using a can of compressed air before inserting it into the scanner. (Hold the can upright and give it a test squirt if new, otherwise it will leave a gas residue.) Then give it a wipe with an Ilford (or similar) anti-static cloth kept clean especially for this purpose. With these two steps, you should hardly need to do any clean-up of the scan. (Even if your scanner has ICE, it doesn’t work on all emulsions and some people claim it softens the image so it’s better to get the source material as clean as possible.)
  2. Scan at 48 bits per pixel, maximum resolution with maximum cropping, save as 48 bit TIFF and reduce the TIFF size by 2 unless you’re aiming for a large print. From a 2700 dpi scan, this gives a 1350 dpi file that will be better quality than a 1350 dpi scan and isn’t too large for manipulation in less than half a gig of RAM (while still good enough for an 8x6 print). Using VueScan, I save a raw scan file and the normal processed ‘crop’ file; I usually renumber them to match the frame numbers and put them in a subdirectory named after the roll number or code.
    Don’t do any heavy manipulation in the scanning program (too fiddly), other than some slight curve or end point adjustments to the histogram.
  3. Load the processed file into the image editor. Crop out any black edges outside the frame and (if required), straighten the image by tilting then crop again to straighten the edges.
  4. Go to ‘Levels’ and try the ‘Auto’ adjustment. Usually this, and perhaps sliding the midpoint around to lighten or darken the image as desired, is enough. In odd cases where the exposure is off and Auto causes radical colour alterations, clip one or both ends of the histogram manually in each channel.
  5. If grain aliasing is present (generally with 200 ASA film and up), use the despeckle filter with default settings. This will slightly blur the image but it usually cleans up the aliasing. It can also be used on selected parts of the image to clean up small dust specs and marks, or on sky areas that show the grain more. Despeckle sometimes causes posterisation; in such cases, try Gaussian Blur with a very low value.
  6. Use the clone tool with a small, soft brush to paint out any larger dust marks or hairs (or unwanted objects, reflections, relatives, …).
  7. You may want to fix up the colours at this point, as minor casts aren’t unknown. Unfortunately, my ability to distinguish finely between different visual tones is about as good as my ability to do the same thing with musical ones (there’s a reason I play drums and only my ability to keep time hampers that). Hence I rarely bother, providing the image looks natural to the untrained eye (like mine). If your exposure is correct then auto-levels usually removes any cast. Otherwise, try the colour picker on regions that should be grey and check that the RGB levels are roughly equal.
  8. Finally, apply Unsharp Mask: radius=1; amount=0.5-1 (or 50-100% in Photoshop); threshold=1-6. (If you can, try to make a selection of the sky area, invert it and only apply the unsharp mask to this, which will avoid sharpening grain in the areas most likely to highlight it.)
  9. Save the image as a compressed TIFF with the same frame number and a different name (e.g. “gimp0012.tif”).
  10. For web posting, scale the image down to 500-600 pixels on the longest side. This will soften the image so apply a touch of sharpening again (1; 0.3/30%; 1); missing this step is the main reason why many web scans look soft. Boost colour saturation by 10-15% for most images except Velvia scans or polarised shots. Save as JPEG with the most quality you can get away with for a file size under 70Kb (typically 85-90%) and a small amount of smoothing (1-2%).

I did say it wasn’t perfect. However, using this procedure I can churn through a dozen scans in a couple of hours, and it does at least provide a starting point for scans that don’t make you wonder if you’ve wasted your money.