Friday, February 20, 2009

Digitizing slides

In the days of film cameras, one of the ways to take a lot of pictures cheaply was to use slide film rather than standard film. The film was around the same price, but when developing slides, you didn't get any prints, so rather than a $10 or $15 bill for the developing, the bill was just $3 or so (if I remember correctly -- in any case it was way cheaper). Others also would claim that the slide film was better for pictures & sharing since you could no project them to an audience (back in the days of 9 and 11 *inch* black and white TVs and *no* computers, there wasn't any other way to do it).

So, I have thousands of slides that I have taken over the years(several hundred from my honeymoon alone) and my mother-in-law brought over a bunch of slides that Angie's father had taken over the years (going back to the late 50s). I want to get all of these scanned into the computer so that we can share them and, if desired, print them.

Before I get into the nitty gritty, I want to lay out some ground rules that I have for scanning large batches of slides/negatives. These have grown out of my experience scanning film and your mileage may vary, but I think they are a good starting point for anybody thinking about a similar project. They include:

  • I want the process as automated as possible so that I can do real work while the scanning is going on. Processes that require manual intervention every few minutes means that I have to dedicate larges amounts of spare time that I just don't have (like any of you do).
  • I want "good enough" quality pictures to come out of the scan process so that I don't have to do any manual processing of the photos (other than rotating them). When I first started scanning negatives, I would do a raw scan at high resolution and then spend 15 to 20 minutes per photo to get them to a state where I liked them. This is clearly unacceptable for large amounts of photos.

    So my model is to get them good enough off the scanner so that I can enjoy/share/watch/etc. without any manual processing.

  • I want to be able to easily figure out which slide/negative the photo came from after I'm done scanning in case there's a picture that I want to do more with (such as scanning at high resolution and lots of manual processing so we can print out an 8x10 or 16x20 photo). This means that I need to be able to figure out which negative from without having to resort to a manual search of thousands of slides.
  • I want to preserve the film in case someone wants to work with it years from now.
  • Speed is not the driving factor. Scanning thousands of slides/negatives will take time. What is key is that the work can be done while I'm doing other stuff. This leads to some choices on the scanning which actually make the scans take longer, but you get better quality scans and you get to keep working on the day job while you're doing the scanning.

These ground rules led to a number of choices I made in setting up this process. As I describe the process, I'll try to explain why and how I made these choices.

Choosing the scanner

The first issue to address is how am I going to scan slides themselves. There are two basic options for scanning slides:

  • Using the slide adaptor that comes with most flatbed photo scanners (if you have a multi-function device (otherwise known as an all-in-one), you're probably out of luck as they don't seem to come with options for scanning slides). These adapters typically require that you place some number of slides (typically 3 or 4) into the adapter, remove the typical white background for document scanning and then scan the slides).

    I find this process painful for many reasons, the biggest one being that it's very time consuming and manual in nature. However, this isn't too bad if you don't have a bazillion slides to process.

  • Using a film scanner designed to scan slides and negatives (film) rather than scanning documents/photos. These typically do a much better job on film that the flatbed scanners and they usually also have substantial automation capabilities.

It just so happens that I have both types of scanners and for me the clear choice was to use the film scanner. My film scanner is a Nikon Super Coolscan 4000 ED (it's about 5 years old and has been superseded by the newer 5000ED).

Organizing for scanning

If you're like most people, your slides have not stayed in their little boxes that you get back from the developer and frequently they are intermingled (in some cases within one of those slide projector trays, in other cases in the little slide shoe box where you threw all the slides).

One note about handling slides: Most slides are raw film stored within a cardboard or plastic mount which just holds the film without providing any protection to the film itself. You should use care when handling the slides to keep fingerprints, water, dust, etc. off the slides. I recommend using low-cost lint free gloves available at most photo shops when handling the slides.

You can choose to stay with the disorganization and just scan things, or you can put the slides back into their original sets. I chose to do the latter because figuring out what's on slides and telling stories about them frequently his helped by the nearby slides on the same strip of film. Getting the slides back into the set and then perusing them in order helps greatly.

To get them back into sets, you need to look at each slide. Most slides, even those printed many years ago, will have two pieces of information on each slide. A slide number in one of the corners and a processing month/year stamp. Sometimes this information is printed on the slide. Sometimes it's embossed in the cardboard mount. In many cases, the printing is hard to read and you have to use some sleuthing to figure out what set the slide belongs to and what slide number it is in that set. In the slide below you can fairly clearly see the slide number (34), but the processing date (May 89) is embossed on the cardboard and a bit harder to see.

Once I had them all grouped in sets & ordered by slide number I simply rubber banded them and put them into my to-be-done box and then started cranking.

Scanning the slides

Setting up the scanner

My 4000ED has an optional slide feeder (SF-200) which can feed up to 50 slides at a time for automated processing. This is ideal for my project. However, in many of the reviews of the product and in various support web sites, I found that there were many complaints about slides jamming in the machine -- which would really interfere with my automatic process requirement. I came close to just blindly upgrading to the latest version of the feeder (SF-210) thinking that it had to be better than the one I already had. However, from the reviews that didn't seem to be the case.

I should note that after looking at the wide variety of slides that I had in my collection (especially when I added in the older slides from my mother-in-law) it isn't so surprising that this is an issue. The slides vary greatly in materials (plastic, cardboard, even some metal) and they varied greatly in thickness.

All that said, I found one suggestion in an Amazon review that recommended tilting the scanner about 10 degrees and instead of using the spring-loaded slide pusher, place a C battery into the tray (it would roll down with the slides adding just a small amount of continuous, even, pressure). I gave that solution a whirl and across about 2K slides only had 6 or so jams -- two of which were caused by material defects in the slide mounting (the film had curved out of the mount and caught on the next slide causing the two to load simultaneously). Not bad.

To accomplish this I used two index card packs to raise the one side of the scanner and just placed the battery into the tray as you can see below:

Setting up the scanner software

Nikon Scan 4 is the software package that comes with the scanner. I modified the default settings to enable the following features:

  • Enabled Digital ICE - which does a great job getting rid of dust and small scratches -- it's not perfect, but it does work pretty well.
  • Enabled Digital ROC and Digital GEM post processing - these do a level of fade & color correction that makes many scans presentable that otherwise wouldn't be without a lot of manual processing.
  • Enable multi-scanning 2x - each slide is scanned twice and the scanned data is averaged together -- this gets a better scanned picture on most slides.
  • Set resolution to 2,000 pixels/inch (about 1/2 the full res quality of the scanner) at 100% scale. Just to keep the pictures down to a reasonable size on disk and to make some of the post processing more efficient. I can always come back later if I want a better quality scan on a particular slide.
  • For each batch scan, I set the file name to a one up sequence starting with the year (so, for example, the slides I recently scanned had a base file name of si2009001 and a two digit sequential number of the slide within the slide set). When I processed the next batch, I would increase the base file name by one (e.g. si2009002). The net result is that I could tell which slide set and which slide within a slide set a digital file came from . For example, a digital file with the name si200904523.jpg came from slide 23 in the 45th slide set scanned in 2009.

Loading the slides

Emulsion side - Each slide has an emulsion side and a smooth side. The emulsion slide is the side that the image is recorded and it recorded backwards (to view the slide correctly you view through the slide from the non-emulsion side. This is important because most scanners will tell you that they want the emulsion side facing a particular way (either by directly mentioning the emulsion side, or by using pictures of a slide with an ABC on it (when ABC is backwards you are looking at the emulsion side). On most slides that have some kind of printing, the side that indicates "this side toward screen" or something like that is the emulsion side and the slide number and date stamp are typically on the viewing (non-emulsion) side.

Up vs down - the orientation of the slides (which edge is up) seems to be somewhat random with respect to the printing on the slides. In some cases they are both in sync (the slide correctly oriented when the number/time stamp are on the top. In other cases it's the opposite (the number/date stamp needs to be upside down on the bottom in order for the slide to be oriented correctly). I found I had to look at a few slides to figure out which way it worked with that set.

Landscape vs Portrait - while slides usually appear square, the film within the slide is not. When you're holding the camera horizontally (the normal position) the image will be recorded in a landscape mode (where the width of the image is longer than the height of the image). When you're holding the camera vertically (on its side) the image will be recorded in portrait mode (longer height, shorter width). This is important in slides because in most scanners you should not turn the slide to correctly orient the picture if it was taken in portrait mode. Just scan the picture in landscape mode and later, in software, rotate it 90 degrees to get it into portrait mode. The reason for this is that most scanners only scan the landscape portion of the slide and will miss some of the slide while recording some of the mount if you scan the slide in portrait mode.

Slide Numbers - most slide sets do not start with slide 1 (at least most of mine did not) and frequently that have slides missing (sometimes simply because the slide image was blank). I wanted the actual slide numbers to match the file names so I would start the file numbers with the first slide number and I would ensure that all slides were sequentially in order, filling in missing slides with slides from the end. When I had to do filling in, I would go back to the files after the set was scanned and manually renumber the fill-in slides to correctly represent their slide number.

Scanning the slides

I would simply load a set into the feeder (correctly oriented, emulsion side to the right when looking at the scanner) indicate in the software that I was feeding X slides and set the starting number at Y. Then I was off to do the real work while the scanner went along chugging through the slides in the feeder.

Slide Storage

In order to be able to quickly locate slides, as well as to provide for archival storage of the slides, I chose to use Print File Archival Slide Preserver sheets for the slides and placed a label on each sheet indicating the slide set (which was part of the digital file name) that the sheet contained:

You can get these at many photography supply stores. I purchased my at Archival USA.

Once I had the slides stored in the sheets, I placed the slide preserver sheets into Century Box Archival Storage Albums (that I also purchased from Archival USA). Another option would have been to buy the file hangers that Print File makes and simply hang the sheets in a file cabinet, but I preferred the storage box. Anyway, I placed the slide pages into the boxes and placed labels onto the boxes indicating which slide set ranges were in the box.

Miscellaneous Tidbits

Use the magnifying glass, Luke

I found having a magnifying glass quite useful in trying to determine the slide numbers and/or date stamps on slides as well as to try to determine the orientation of the slides on slides that had no markings. It was just plain useful. Get one and have it nearby when you're working on the slides.

Remounting Slides

In some cases, it might be worthwhile to remount slides. For example if the mount is damaged, too thick, or otherwise interferes with being able to scan the image. I had this with one particular set of slides that came from my mother-in-law. It seems that in the late 1950s in Europe, slides were mounted in metal mounts that sandwiched the film between two pieces of glass. When they got to me, they were in pretty sad shape:

So I ordered some slide mounts and peeled back the metal cover, separated out the film from the glass sandwich and mounted them into new slides which scanned much better than the originals had.


This process seems long and arduous, but in reality the most time consuming part (other than the remounting of that one metal set) was the organizing the slides step because many of the slides were mixed together, some had no writing on them whatsoever, many had slide numbers and date stamps that were almost unreadable (magnifying glass helped there sometimes).

Once the scanning got started, the process essentially amounted to about 5 to 7 minutes to swap slides and store the scanned slides every hour an a half or so (that's about how long it took to go through the average 30 or so slides per set with the settings I had used on the scanner software).

I'm very happy with most of the pictures and for those that I'm not happy with, the slide itself usually left a lot to be desired -- almost always because of low exposure on the film.

Tags : / / / / /


Anonymous said...

Thank you for this post.

If you also know how to digitize the old kind of home movies you needed a projector to show, I'd love to know about it. :)


id scanner said...

Great info on scanning Technics and machinery. I see you're a fan of Nikon when it comes to scanning :)