Flatbed Scanners -- or "Am I an Idiot?"

Okay... Just had one of those "Ah-hah!" moments. I was comparing scanners today at my local photo store -- Epson 4990, MT i900 and MT i800.

First off, the Epson was fastest, the i800 sharpest by a touch and slowest by like 3x, and the i900 the softest and slower than the Epson. The "new" holder on the i800 is no great shakes. You need four hands to hold your tranny in place while you lock it to avoind crinkling the tranny. Add the incredibly slow scan times and this is one scanner to avoid if you plan on scanning 4x5 at any significant resolution. The i900 was softer and slower than the Epson and had a weird halo on high-contrast edges. (So guess which scanner I brought home...)

Anyway, back to my stupid... I was testing these scanners out on a piece of 4x5 tranny. I used them all and came back to my office to compre their results to my aging Epson 3200 pro -- a scanner I've had a love/hate relationship with for the past few years. Great for casual reflective scans, marginal for tranny and really sucked for 4x5 since the holder was crap and I always got Newton rings. Here's when the "Ah-hah" hit. I am SURE somebody else has realized this and I simply missed the memo... But in my defense I have been a sort of 3-year hiatus from the 4x5 film world.

But now I'm back. So in comparing my 3200 scans to the above breed, it's age is clearly showing. Even with Silverfast, the best I could manage was still softer and had less DR than any of the above newer scanners and itr is slow too. No surprise. As I put my test tranny in the crappy holder I asked myself why a company like Epson could not design a better holder -- FTR, the holder for the 4990 is essentially unchanged from the 3200, only there are 2 slots so you can scan 2 4x5's at once. Wow.

The Epson manual clearly states to load the base side down with the tranny inverted left/right. Base side down with 4x5 means the smooth shiny side of the tranny -- also the side that bows out -- can also easily contact the glass surface of the scanner and generate Newton rings. Dang, no free lunch... No problem, I placed little shims on my 3200 holder to hold the bowing tranny off the glass, I could do the same again with the 4990. Or I could buy that guy's fancy holders...

That's when I decided WTF, why not load the tranny emulsion side down? Now the tranny bows away from the glass the way it should and then gravity works for me, drawing the center of the tranny almost perfectly flat, yet still leaving an air-gap above the glass. Sure the image would be reversed, but a mirror-flip in CS is a piece of cake. So I tried it. Result? Best scan I ever got off my 3200, period! Ran back to camera store with test tranny, loaded it upsidedown in the 4990 holder and scanned. WOWSERS! WAYYYYYY sharper than the first set of scans with the tranny loaded -- errr -- correctly... So I came home with the 4990.

Okay, certainly somebody is going to tell me this is old news and ask where the ____ have I been...

Feeling stupid in Silicon Valley,

Thanks for the very informative comparison, Jack. Those are the models I'll be looking at in a month or two.

I hope you'll keep us updated on your experiences with the Epson.

Good thing I never read the instructions. I just thought the emulsion should be close to the glass. Now you've done it-- triggered lust for a "new silver bullet". Have you tried a 8 x 10 neg? thanks for the report. Jim

Hi Jim:

Re 8x10 scanning, Epson supplies what appears to be a sheet of OHT film for scanning up to 8x10 on. It is not a holder per se and I suspect it is some form of anti-Newton plastic since it rests directly on the scanner glass and the tranny/neg in turn sits directly on top of it.

I don't shoot 8x10, but would be happy to do a test scan of 4x5 resting on top of this sheet and post the results here. I am busy for the next few days, so it will probably be next week before I get the new scanner up and running and can do this.

Cheers,

Jack

Scanning with the emulsion away from the sensor always seemed crazy to me, and I assumed no one actually did that. It certainly makes for a softer scan on the 9550. That'l teach you - ignore those instructions.

Hmmm... I own a Microtek i900 and am happy with it, although I've never tried the Epson 4990. The only instances of "halos" I've seen were related to sharply ramping a curve (overexposing) a flat negative, which in my experience causes those halos to appear in many flatbed scanners.

i'm guessing that the improvement is coming from better focus, not specifically from which side of the film faces the glass ... the epson's have been known to have a lot of variance in the focal plane. if the neg is bowing visibly one way and then the other, you might be changing the focal plane by well over a millimiter. people who wet mount use shims to find the plane of best focus (and fine tune by flipping the emulsion side one way or the other).

I tested my 4990 and found sparpest with B&W negs emulsion up

The 'guide' for 10x8 is just that, a guide. The internal dimensions are greater than 10 x 8, so you can move the film about. The film just lies on top of the glass, which means you can get newton rings whichever way up you place the film.

My problem with scanning through the emulsion is that it is not perfectly clear and must add some diffusion to the data. If it is necessary for focus, you might try a shim of paper under the carrier instead. Then you could get the emulsion facing the sensor and be in focus. I found a shim of folder paper worked fine to raise my negatives enough to avoid newton's rings, and did not affect sharpness. The 9950 does seem to be focused just a bit above the glass, which makes sense - while you can scan flat art at hig rez, there is not much flat art that is high rez so who would know if it is a little out of focus?

I've never had trouble with the Epson 3200 film holders as far as Newton rings are concerned. I've also made my own film holder which doens't cut off as much on the edges, but also doesn't grip the film securely. It also doesn't produce newton's rings. It may be that for some reason my film is more rigid.

The last time I tried the film both sides up and compared, I couldn't tell the difference. Perhaps I should try again.

I've always assumed that Epson tells you to keep the emulsion side up so the resulting scan won't reverse left and right, not to get a better scan. I believe the scanning software is designed to work with reflective scanning, and without a reverse built in for film, it would normally reverse left and right if the base side were up. Of course, they could have implemented a reversal in the firmware or software when doing film, but they decided not to for some reason. If anyone has any further light to add, please do.

>>The 'guide' for 10x8 is just that, a guide. <<

I just opened my 4990 and yes it is just a guide -- My bad.

There was an 8x10 "holder" ike I described at the store with a film surface -- it must have been for some othter scanner. Sorry for any confusion.

I tested my 4990 and found the scans are slightly sharper emulsion side up.

Right you are, Jack! I had exactly the same issue with my 2450: Newton's rings. Loading the trannies emulsion side down seems to keep them from sagging onto the glass, at least during the time I need to scan, and I find the sharpness as good as before, or changed so little as to be no matter.

On a related note, do these scanners have the ability to change focus? I seem to recall that some scanning software will do this with certain scanners. Vuescan?

Well it seems the camps are divided on this subject... To satisfy my own curiosity, I tried a different tranny -- flatter but still a tranny -- and saw virtually no difference scanning emulsion up versus emulsion down. Soooooo... I can only conclude the direction makes a difference sometimes. Use what suits you.

FTR, I was scanning Provia and Velvia using Silverfast with auto-sharpen set to medium and off, no ICE, no scratch removal, and single-pass scans only. Obviously other film types and other setting combinations could produce different results.

I did experiment with scan resolution settings and found the following with two different transparencies -- and this was to my eye only so I share this only as an indication and not an absolute: I compared 2400 DPI, 3200 DPI and 4800 DPI, and found that the 2400 DPI scans looked the sharpest, followed by the the 3200 which was slightly but notably softer, and the 4800 DPI setting softer still. I even tried the scan at 4800 DPI and resize to 2400 DPI routine and found -- again to my eye only -- that the base scan at 2400 remained sharper.

All offered FWIW only and clearly YMMV!

Cheers,

I've scanned both ways with the 4990 and never had Newton rings either way. I kind of like the holder, it's a whole lot better than the flimsy piece of crap that came with my Linoscan 1400 and that I had to hold together with electrical tape because it started cracking after a year or so of use. But back to the topic. I've always assumed that Epson recommends base side down so that the image will be correctly displayed on the monitor, not because it had anything to do with Newton rings. However, I do scan with emulsion side down because I've read and heard from people who are supposed to know that the image will be a little sharper that way. It's easy enough to invert the image in Silverfast or Photoshop.

Not to toot my own horn here but you guys need to read these posts more. This is old news on the Epsons. I posted this same thing many times starting about 2 years ago about reversing the neg. and flipping it in PS to avoid the sinking middle phenomena. But you knot what works even better? Tape the lid shut and tilt the scanner on its left side to take the weight off the center of the film. The light rides on a rail on the right side. I estimate about a 5% sharper scan. Works like a charm.

I think I referenced that trick of yours in a post recently, Kirk. The budget-priced mechanics of these things probably need all the help they can get, and you can in part make up for some of the cheap plastic by understanding their shortcomings and finding ways to compensate. Let's hope, anyway(!)

Jack,

I just scanned an image twice, once normal and once flipped (single pass scan). Sure as hell, the flipped version is sharper, in the center and the edges. Thanks for the tip. I will have to try Kirk's suggestion and flip the scanner on its side.

Dunno if you're an idiot, but...

Buy one of Doug Fisher's MF film holders. Not perfect, but makes scanning _lots_ easier.

jbh

Jack, could you advise me how you are able to set up your 4x5 chrome on the 3200 and actually get a scan of a 4x5. I have tried repeatedly, but with no success. Your assistance would be appreciated. Thanks. JW

>>Jack, could you advise me how you are able to set up your 4x5 chrome on the 3200 and actually get a scan of a 4x5. I have tried repeatedly, but with no success. Your assistance would be appreciated. Thanks. JW<<

Well, I don't have it hooked up any more so I cannot give you screenshots, but...

My 3200 was the "pro" version and came with Silverfast and a holder for 4x5 -- I just set the 4x5 tranny in the holder, selected transparent and positive from the software dropdowns and scanned.