A Scanning Workflow, Part 3: Scanner Settings

Posted on 25 June 2010 by Guest Contributor

This guest article was written by Sam Agnew. You can find Sam on Flickr or on his personal photography blog: Smash and Grab Photography. Also visit part 1 (making choices) and part 2 (film preparations) of this series.

OK, let’s put that clean film in the holder. This is where you want to take care to align the frame edges carefully with the edges of the holder or you will lose parts of your image. Also, don’t touch any part of the visible part of the film with bare fingers. Scanned finger prints are obvious and HUGE.

Read your scanner’s manual about how the film should be placed. Film has two sides. The shiny side is the backing side and is harder and more scratch-resistant than the other side. The duller side is the emulsion side. It has the actual layers of chemicals that form the image and it was the side that was facing the lens when it was inside the camera. Most scanners want you to place the film in emulsion side down but NOT ALL. Putting the film in the wrong way up is most likely to affect image sharpness. It may also lead to you having to flip the images left-to-right.

Resolution

If you are scanning for maximum quality you probably want to use the highest resolution available. Most modern films will show considerably more image detail at 4000dpi than 2000dpi. These two settings correspond to roughly 24 megapixel and 6 megapixel for 35mm film. 35mm gets enlarged so much for normal sizes that you want all the resolution you can get. However, keep in mind that if your scanner is a flatbed or otherwise poor optically you may see no more detail in the image at higher dpi.

Dot Dot Dot
Creative Commons License photo credit: yago1.com

Basically, the problem with a lot of flatbeds is that the “dots” seen by the scanner at high dpi settings have a great deal of overlap with each other such that not much more detail is seen even though the files are getting larger and larger. Additionally, they are often fixed-focus with a thin depth of field and the optics are often of a middling quality. On many flatbeds it is pointless to scan at more than 1500dpi or so because the image they produce is so fuzzy and lacking in definition. The Nikon Coolscan scanners, most of the newer Konica Minolta scanners and of course the big Flextights will not have this problem and you will see correspondingly more detail all the way up to the scanner’s maximum resolution.

On the other hand, increasing the resolution only increases image detail at larger magnifications or print sizes. At a web size, there may be no difference at all from the higher res scan except the loss of four times as much disk space. Resolution is the only setting that will be invisible at smaller sizes.

Here is another consideration: Each doubling of dpi represents a quadrupling of file size. A 2000dpi scan is one quarter the file size of a 4000dpi scan. This is why current flatbeds are such a terrible waste of bytes. If each digital pixel is not conveying a discrete quantum of image detail then it is quite simply a waste of electrons. Unnecessary pixels will slow your workflow, fill your disk and give you absolutely no benefit unless you are printing or displaying at large sizes.

One last time, resolution is almost the only image parameter you control which makes no difference to how the image will appear in web size or thumbnail.

Bit Depth

Most good scanners will offer a choice of 16 bit or 8 bit scans. Your computer screen will only show you 8 bit colour so one argument goes that scanning at a higher bit depth is pointless. However, if you plan to do any adjustments to the scan after scanning then the greater detail in the 16 bit file may allow you to alter contrast, brightness, saturation and colour balance to a greater extent without degrading the image. In contrast to dpi, the difference in file size from 8bit to 16bit is merely a doubling.

A similar argument relates to Colour Space. There are some wider colour spaces available which have fans and detractors. On the plus side, these larger colour spaces can theoretically show more colours. On the minus side, things will look wrong if the output device or application fails to recognise the colour space. Also, because the same number of bits are spread through a wider colour space there are less gradations available in a given colour.

When I’m scanning at 16 bits I have more than enough bits for any colour space. Additionally, there is only one application which I have to worry about recognising that space and that is my image editing application. For this reason I always scan at 16 bit into the Wide Gamut RGB colour space. As a general rule, try to get your source image at high bitrate in a large colour space but expect to output into a small colour space (sRGB, usually) at 8bit per pixel.

Format

TIFF or Adobe DNG. The format itself is not that important but I would choose one that allows 16 bits and that isn’t “lossy” like GIF or JPEG. TIFF compression, if used, is lossless if done using the common “LZW” compression so don’t worry about LZW-compressed TIFFs. Your display cannot show 16bit per channel colour depth but it will give you much more latitude for exposure and colour correction in post.

If you know that you will not be editing your scans further then by all means output 8bit, JPEG, sRGB. I would always make my final file (after all editing) an sRGB JPEG unless you know you have a very good reason not to.

ICE

Cauliflower ice
Creative Commons License photo credit: schoeband

On some scanners ICE works better than others but on EVERY scanner it works better than any purely software dust removal scheme. Cloning out dust and scratches by hand will quickly drive you around the bend. If your current scanner doesn’t have ICE then start planning to replace it unless you are capable of keeping your film scratch-free and squeeky clean. If your scanner has bad ICE that degrades the image you should also plan on replacing it.

ICE is actually a tradename but (like Kleenex and Hoover) has come to be widely used to refer to any system that uses scanner hardware to automatically fix dust and scratches. The principal relies upon most colour film being transparent to infrared. By scanning in RGB and Infrared, the Infrared channel can be used to identify dust and scratches and remove them. The current Nikon scanners do this to an almost magical degree. Canon’s system (called “FARE”) is also well-regarded though not quite as good. Epson also licence ICE for many of their scanners. Unless you operate in (and get your film developed in) a clean-room environment the time saved by good ICE is almost incalculable.

There are two general exceptions to this magic technology. One has an answer, one doesn’t.

For Kodachrome the answer has traditionally been that ICE won’t work. However, the Nikon Coolscan 9000 ED has an advnanced “ICE4″ system that works with Kodachrome images.

For black and white the answer is still no. Traditional black and white film is not transparent to infrared so the sytem does not work on that type of film. Attempting to use it anyways will probably yeild strange results.

The problem with software dust and scratch removal is that once an image is a digital RGB picture there is no difference between dust and a picture of dust. The ICE system can use the infrared scan of the film to tell the difference.

Other Settings

Most scanning software has tons of settings which aim to allow you to (eventually) produce a print-ready image straight from the scanner. Working this way is very slow and frustrating. My advice is to ignore all such settings. You want a scan with reasonably close colour accuracy and no clipping but don’t bother trying to adjust curves or anything like that in the scanning software. Take care of all of that in post.

One exception to this rule is if your scanner does poor scans of negative film. Some scanner software is better than others.

If your scanner software is bad at converting negatives to positive you may want to manipulate the scanner gain for each primary colour to achieve a “true” negative image by scanning your negative film as a positive and using an image processing program to convert the negative. See this fascinating writeup: http://cjeastwd.blogspot.com/2009/04/negative-film-scans-on-nikon-coolscans.html. My experience with the Coolscan 9000 was that it did a good job of scanning negs right out of the box.

If you are going that route then I would highly recommend trying a plugin called ColorPerfect (previously, ColorNeg). It seems to do the most natural job of creating positive images from raw positive scans. I am currently using this with my cheap Konica Minolta scanner and getting better results than I used to get from either Konica Minolta’s software or Vuescan. Their main English webpage is here: http://www.c-f-systems.com/Plug-ins.html

7 Comments For This Post

  1. Janne Says:

    Worth noting: for whatever reason I have a lot less problem with dust on medium format film than 35mm. Of course, each speck of dust is larger on a 35mm frame but it goes well beyond that. There’s just a lot less dust on MF for me, for some reason. I never bother with ICE for MF scans; it’s faster for me to go over the image with a healing brush than spend the time to do a separate IR scan, and I generally want to look over the image in some detail anyway.

    [Reply]

  2. Thomas Says:

    I use a piece of cloth with which you can clean your glasses. Just wipe your negative before you put them in de scanner. Dust glues it self to those cloths, works really well.

    [Reply]

  3. K. Praslowicz Says:

    The day I remembered that my scanner has ICE and turned it on is the same day I started to really love shooting color film.

    [Reply]

  4. Paula Says:

    Just wanted to thank you for writing this up. This is a great resource for me as I undertake a big summer project of scanning my negatives. I’m still really fuzzy on resolutions and target size v. document size, but this has really helped significantly narrow down my confusion. Many, many thanks

    [Reply]

  5. Sam Agnew Says:

    One of the ways that a lot of scanner software causes confusion in an attempt to be helpful is when they ask you to specify resolution in terms of output dpi at a certain print size. You always want to know what actual dpi the scanner will perform the scanner because you always want to scan at the native resolution or half or one quarter and so on. For example, if your scanner is 4000dpi you will get the best results at 4000dpi or 2000dpi or 1000dpi or 500dpi. You don’t want to ask the scanner software to interpolate.

    Try and see if your scanner software will let you specify a raw scan dpi rather than asking you to specify some combination of target print size and print dpi. If you need an intermediate value I would just choose the smallest resolution that gives you what you need even if it is giving you a little more than you need.

    Also, as I will get to in a later article, you should experiment to see what resolution you really need. I’ve gotten some excellent prints done at quite a bit lower than the “magic” 300dpi print value with good images scanned by a good (Coolscan) scanner. They don’t show any lack of detail even when looked at quite closely.

    [Reply]

  6. jojonas Says:

    oops! this kind of got me confused. I shoot mf and was planning to buy either epson v600 or canon 9000f.
    also, all this talk about nikon coolscans… I thought they were out of production?

    [Reply]

  7. Sam Agnew Says:

    I just bought myself my own Nikon Coolscan 9000ED. I bought it brand new and I’ve had it less than a month.

    Nikon have long kept special items available for long periods (most of their “legendary” lenses, for example) and produced them in small quantities to order. It seems the current Coolscans (5000 and 9000) are now in this category. Order and you will receive! You just may have to wait a little.

    [Reply]

Leave a Reply

Advertise Here

Photos from our Flickr group

See all photos

Advertise Here