Scanning and preparing photos for internet

by Vadim Makarov for his Photo Pages

Most of my slides are scanned on a Nikon LS-2000 dedicated film scanner. I scan in full optical resolution (2700 dpi for this scanner), full dynamic range and no automatic adjustments during the scanning stage. The raw images are saved in a lossless format (e.g., TIFF). This way, full optical information is extracted from the slide and saved at once, with no need to return to the scanning for possible later image uses. I run the original Nikon software for scanning (Nikon Scan 3.1.2).

After that, I play quite a lot in Photoshop with curves and hue/saturation adjustment layers on each image individually, to achieve the best look. This overall tonal and color adjustment is the most time-consuming step. In my experience, it is the key for getting nice-looking web images. Unfortunately it’s hard to describe it. To me, this has been a skill acquired through practice. The adjusted images are saved and archived with adjustment layers in .psd format. (When tonal- and color-adjusting, you may think the image is ready... stop and wait a few days, then look at it again. You’ll see what final touch is needed. Sometimes I even have to return and improve adjustments a few days after the pictures are added to my website.) I am really sorry I can’t describe this step in detail: I just don’t know how to teach it.

The scanning and adjustment steps are done only once for each slide. Then, images for any final use (e.g., commercial use if asked for) can be generated from the archived Photoshop file.

I generate final JPEGs for the web in several sizes, setting Unsharp Mask parameters for each image size manually, much as described by Philip Greenspun. If you don’t have time, this step could be automated with a fixed set of Unsharp Mask parameters. However, I always end up using different parameters for different images. So, I like to keep it manual. Here are my starting points for each resolution:

Image size, pixels 2592x3888 1024x1536 512x768 256x384 170x255 128x192
Amount 80 90 90 (actual Amount set for 512x768)−5 100 100
Radius 1.2 .. 1.6 (depending on original image sharpness: use narrower radius for sharper images) 0.6 0.5 0.5 0.4 0.3
Threshold 4 .. 6 (depending on noise/grain level: you don’t want to sharpen noise) 2 1 1 1 1

When saving, I set JPEG quality level in each resolution so that no details are visibly distorted, or there is a slight visible deterioration at worst. I especially watch for red-colored details (red channel is most compressed in JPEG) and people’s faces. I err on the side of higher quality and larger file size, reckoning that bandwidth is becoming progressively less of an issue. I usually end up setting JPEG quality in the 7–10 range (on the 0–12 scale of Photoshop CS3).

Until recently, I haven’t been color-managing through the process. I scanned and edited with color management switched off, preparing images to look fine on my monitor (after having checked that my monitor displays a full tonal range, has a gamma around 2.0, and setting its color temperature at 6500 K). Now I’ve begun using color management through the scanning and editing. I’ll see what comes out of it. Anyway, I have to convert images to sRGB color space before generating final JPEGs. As of 2009, web browsers do not color manage, so I have to aim for a range of screens out there. Many screens are still close to sRGB (most laptops are), but at the same time an image shoudn’t look very garish on a wide-gamut monitor when displayed in a browser. Ideally you should view your web images on as many monitors as possible, to get a sense of what they can look like on different screens, and aim for the middle.

One thing I’ve learned about film scanners is the importance of cleaniness. ISO 100 color slide film has a greatly amplified contrast range (1:10000 optical density range for a four-stop, 1:32 scene contrast range). At this contrast range, any dust in scanner optics produces unpleasant artefacts. I keep my scanner in a plastic bag when it’s not in use.

I scan slides mounted in anti-Newton glass frames (Gepe VR). This keeps the film flat. Thus, focus across the frame is usually not a problem. Anti-Newton glass adds some extra noise to the image, appearing as small-scale low-contrast grain on the scan. This is visible in the two largest web image sizes (especially when the photo contains smooth areas like sky, and you Unsharp Mask it). However, this increased noise has never been a problem in a large-size print reproduction the few times my images were licensed for commercial use.

Besides the Nikon LS-2000 and LS-30 scanner models, I’ve occasionally used other brands of film scanners. Most of them had poor shadow visibility, despite having the same or better manufacturer specs on optical density as the Nikons. They had various other optical problems as well. I’ve also tried (a long time ago) a Kodak Photo CD lab in Norway and a Russian drum scanning bureau. As the result, I am sceptical of scanning services in general. It is better to have your own film scanner and be in full control of how you work with it, both physically when handling the film, and over the scanner settings.

The Nikon LS-2000 uses an old SCSI interface that can be a bit of trouble to hook up to modern PCs. Several newer Nikon scanners use modern interfaces such as USB. (The newer scanners also have a slightly better optical resolution, but I don’t think that’s important.) Nevertheless, I’ve lately got both Vista and XP desktop machines flawlessly set for scanning, so I see no compelling reason to change my scanner.

In any case, you shouldn’t use a flatbed scanner with a film adapter for scanning film. The flatbeds are not made for the film optical requirements, so the result will suck. Use a dedicated film scanner.


Vadim Makarov
Add a comment | a link