Slide scanners vs. flatbeds

[Ryan McGinnis]

I whipped up this comparison for another fourm I'm on, and I thought that it would be helpful here, as well. Often, when people are considering converting their negatives and slides to digital, they find themselves wondering what's best to purchase to do the job. Many people purchase a flatbed under the assumption that it is the best value -- one can scan batches of negatives and slides as well as photographic prints. They see the DPI figure noted and compare it to the DPI figure displayed by a neg/slide scanner, see the pricetag on the neg/slide scanner, and figure that it'd be foolish to buy the neg/slide scanner.

However, in my experience, flatbed scanners have some very serious limitations when it comes to scanning negatives and slides. The main limitation is resolution -- despite DPI figures, flatbeds are completely schooled by even the lowest-end slide scanner. To show this, I test-scanned a frame of RVP-50 (Fuji Velvia). The photo was shot with a Canon EOS-3 and a Tamron 17-35 2.8-4.0 lens. The negative/slide scanner being compared is a half-broken Nikon LS-2000 (the stepper motor is a bit messed up), and the flatbed scanner being compared is an Epson 10000 XL with the transparency attachment -- altogether, a $3,500 scanner. (It ain't mine, sadly, it belongs to my employer!). You can pick up a used LS-2000 for a couple hundred bucks on Ebay.

Here is a small preview image of the whole slide. The red box that I've drawn shows the area that I'm about to enlarge and compare. I've brightened the 100% crops just a touch to reveal the detail better. No sharpening was done.

[attachmentid=38]

Below is a 100% crop of the area in the box as scanned by the Epson 10000 XL flatbed:

[attachmentid=39]

Last, here is a 100% crop of the area in the box as scanned by the Nikon LS-2000:

[attachmentid=40]

As you can see, there is a HUGE difference between these two scanners. The $250 LS-2000 completely blows away the $3,500 Epson flatbed. Which brings me to my point -- if you wish to scan your negs or slides in so that you can work with them in Photoshop and make prints, then it is very important that you purchase a dedicated neg/slide scanner and not try to get by with a flatbed. I think a flatbed would make okay 5X7 prints and you might be able to squeak by with 8X10 enlargement, but it won't look anywhere close to as good as you will get with even the lowest-end slide scanner.

 

[Dan Robinson]

A slide scanner has been on my list of equipment to eventually get. Have you tried scanning lightning shots with it? In my experience scanners have trouble with lightning because the image is mainly dark with small areas of bright in the lightning channels, and then a blue/purple tint around the illuminated sections of clouds. The scanners I've worked with always seem to have trouble with frames with a lot of totally black areas. The black areas seem to always have noise or lines.

 

[David Wolfson]

The Epson V700 and V750 scanners are due out shortly. They're supposedly dual lens and optimised for pro transparency scanning, and have been getting some pretty good reviews. The V750 also is supposed to have an MSRP of $800. I want to scan up to 4x5 without busting the bank. It sounds like these may finally be in my price range with decent quality.

 

[Greg Campbell]

A slide scanner has been on my list of equipment to eventually get. Have you tried scanning lightning shots with it? In my experience scanners have trouble with lightning because the image is mainly dark with small areas of bright in the lightning channels, and then a blue/purple tint around the illuminated sections of clouds. The scanners I've worked with always seem to have trouble with frames with a lot of totally black areas. The black areas seem to always have noise or lines.
[/b]

"What Ryan said"
Consumer flatbeds just don't cut it for 35mm scanning. They do make an affordable alternative for large format. With the vastly larger negative, you can loose some detail and still come away with an absurd amount of real image information.

Noise is always a problem with desktop scanners when you're trying to dig all the shadow detail out of a slide. I've got an older FS4000 by Canon. It's noise level is decent, although slightly worse than the newer Minolta or Nikon units, particularly those that allow multisampling. (The scanner takes several exposures of a single line, and averages them out.)

FWIW, here's an example of the gyrations I need to make in an effort to to reduce noise and correct scanner inaccuracies.

Here's a typical 'raw' scan produced by the Canon. With lightning images, I need to disable the autoexposure, or else the software blows out the highlights. I then crank up the gain until blocking occurs, then back of a notch.

'Raw' scan

I've given up trying to make a proper profile for this scanner. Since the scanner's response to various films will ....vary, a one-size-fits-all profile is not practical. Instead, I've got a basic profile that gets me sorta close, and a number of complex 3 color curves that take out most of the residual color and luminance inaccuracies. (Beware! A bad profile can clip both highlight and shadow detail - with ruinous results.)

The Canon software is very weak in the color management department. Aftermarket driver packages such as VueScan and Silverfast are supposedly much better. The Nikon and Minolta software is somewhere in the middle.

Rough color/luminance adjustment.

Now, the most important tweak - setting the black point. Any noise below the BP will be buried. Noise above will be reduced. There is no 'proper' BP, only a compromise between a technically 'clean' image and maintaining important low-level details. (It's hard to see at this size, but there is still significant detail in the foreground.)

Some people call this 'photoshopping,' and say it like it's a bad word. Nonsense! You NEED to to this. It is no different than using Ansel's Zone System to calibrate the various aspects of a chemical darkroom. (negative densities, enlarger exposures, paper dev. times, etc.) Whatever imaging system you are using; film, digicam RAW files, vid caps, etc. - you need to define just what zone 0 is represents, at least if you want pure blacks.

In practice, no desktop scanner I've seen will dig all the visual information out of the shadows. They all hit the noise wall before geting all the detail you can see on a projected slide. I've had this and a few other lightning slides scanned by a Minolta 5400 unit. The results are somewhat cleaner, but the difference is not huge.

Got $50+K for a nice drum scanner?

Blackpoint.

Now a look at the slide in on a light table and still more luminance and color channel shape changes, trying to match the two images. Long Fujichrome exposures can really load up on the magenta. (Again, foreground detail is lost at this res.)

Fine Tune.

There is a handy freeware PS plugin that corrects wide angle distortion and vignetting for a number of popular lenses. With my 20mm lens, the effect is subtle.

Distortion / vignetting

After that, I'll throw the picture at Neatimage, to remove the residual chroma noise.

Central cloud detail BEFORE. (600%)

AFTER

This still leaves grain structure and electronic noise in the luminance channel. More Neatimage action, using carefuly constructed profiles, can beat this down to a tolerable level without killing any significant detail.

Whatever scanner you guys buy, be sure to get one with an infrared-based dust cleaning system. Without it, even the cleanest slides will require hours of touch up. Canon's dust buster system works fairly well, but misses small particles when used on the 'low' setting. Since the 'high' setting can occasionally damage fine detail, I'm left with maybe 1/2 hour of dust touch-up per image.


Dan, I'll be happy to scan a dozen or so slides if you want some FS4000 examples.


-Greg

 

[Ryan McGinnis]

A slide scanner has been on my list of equipment to eventually get. Have you tried scanning lightning shots with it? In my experience scanners have trouble with lightning because the image is mainly dark with small areas of bright in the lightning channels, and then a blue/purple tint around the illuminated sections of clouds. The scanners I've worked with always seem to have trouble with frames with a lot of totally black areas. The black areas seem to always have noise or lines.
[/b]

I've never scanned lightning shots before (I don't do much lightning shooting -- I'm askeer'd of getting struck!), but the Nikon scanners give you the option of doing multipass scanning. It takes forever and a day for each slide, but after 16 passes, most of the noise is gone.