Suggestions to Improve Flatbed Scanning of Film - Why I currently hate scanning...

[holmburgers]

I have an Epson 4990, but this line of thinking should work for a number of flatbed scanners with exceptions here and there.

The biggest beef(s) I have with flatbed scanning are the following...

No ability to focus: There is no change in focus between pressing a document flat on the glass or suspending it a few mm's above the glass in a film holder. In either situation, I seriously doubt I'm setting my negatives in the optimal plane of focus.

No exposure compensation: If I have a dense overexposed negative, or a dense underexposed slide, there's nothing but "computational" adjustment of the exposure, leading to noise and artifacts. How hard would it be to take a longer exposure at the CCD or to increase the brightness of the illumination.

Can't scan a whole roll at once: If i can scan a piece of 8x10" film, I sure as shoot should be able to scan a whole roll of 36 exposures!

Film holders inadequate: To say nothing of the overall construction of such scanners (cheap, flimsy, snap-together plastic), the film holders are junk. They're flimsy plastic that is just asking to be broken, they cover up the sprockets and introduce a signficant crop (give me full frame or give me death) and they do an inadequate job of keeping the film flat, particularly along the length of the strip.

Ok... deep breath.....

All told, there's just got to be a better way. The stuff at betterscanning.com looks pretty awesome, but I'm always interested in low-to-no cost solutions, plus their equipment seems to be geared to medium format. The majority of my scanning is 35mm, and that requires the tightest tolerances.

I'm thinking that two 8x10" sheets of anti-newton ring glass would make an idea film holder. This would keep the film absolutely flat and the bottom sheet could be chosen for its thickness; raising the negative stage to the plane of optimal focus for your scanner. This would also allow you to scan a whole roll at once, and in full frame.

I've also given up on my scanners "auto-detect" software to crop out the frames. Not only is it inaccurate and sometimes completely maddening (cropping away the majority of a frame because of a dark background), the scanning mechanism will actually individually scan each frame, reset, and then go to the next. Why it can't just scan the whole platter and then do its business computationally is beyond me. I find that scanning it all in one swoop, and cropping manually is actually faster ("save selection as", or something like that) than letting the software do it for you.

Now, the lack of exposure compensation is tricky, but imagine if you lifted the lid and supplied your own backlight. You could then fine-tune the illumination and peer-through dense negatives/slides much better. Who knows, maybe a good light table would do the trick.

Anyways, just wanted to start a conversation, rant a little, and see if anyone has found suitable work arounds to get the most out of these lowish end scanners.

Cheers!

 

[indigo]

You can try to copy the film using a DSLR.

 

[A. J. Pylvänäinen]

I think what we really need is a new, decent quality dedicated film scanner. I'm really looking forward to reviews of the upcoming Pacific Image PrimeFilm 120. It should be out this summer, but I haven't seen any reviews yet. Hopefully, it'll be good enough to get decent resolution out of both 135 and 120 film.

 

[glhs116]

As far as I am aware the exposure is indeed adjusted when the scanner wants something darker or lighter. Since aperture and illumination are fixed the variable used is time. This is why more fully exposed film takes longer to scan.

 

[williamtheis]

you might also try using Kammi to wet mount under a piece of mylar. very flat. you can get the occasional bubble. I use it on my flatbed. before making investment, try using plain old water first to see if it improves enough. for B&W especially, wet mounting is better since it diminishes the grain. and... no newton rings!

 

[holmburgers]

Thanks guys.

The DSLR option is maybe ok for a single shot, but scanning a roll in that manner? Not an option methinks; plus I don't own one!

That scanner sounds quite nice, but the appeal to me (as a hobbyist) of the Epson 4990 was $$. Very affordable, and versatile in that it can do normal reflective scanning as well. I guess I shouldn't expect to have my cake and eat it too though.

I'm pretty sure that the 4990 doesn't adjust exposure, at least, evidence would suggest that(??). There are no controls in the software for this, and I've not noticed any difference in scan times.

Wet mounting has interested me, but I'm really trying to make this easier as well as better. I am interested in wet-mount darkroom enlarging though. That's how the dye-transfer houses made their separations back in the day; the best quality I understand! But for scanning, I fear that dealing with wet negatives wouldn't improve upon the 'easy' apsect any.

 

[David A. Goldfarb]

The DSLR option is maybe ok for a single shot, but scanning a roll in that manner? Not an option methinks; plus I don't own one!

Can't help you with not owning a DSLR, but scanning a whole roll of 35mm or medium format rollfilm proof-sheet style is very easy with a DSLR, light box, copy stand, a sheet of glass, and an appropriate lens.

A DSLR is also much faster than a scanner of comparable resolution, so say you wanted to scan individual frames, one at a time at the resolution of the camera (i.e., without stitching multiple panels from each frame for higher resolution than the sensor). Unless you have an automated scanner that will feed your film and adjust your exposure for you while you're off working in the darkroom or mixing martinis, this is a much quicker task with a DSLR setup than with a scanner.

 

[jeffreyg]

I have been using an Epson 4870 with Silverfast Studio software for about seven years to scan B&W 120 and 4x5 and occasionally 35mm color and 120 color. That scanner preceded the 4990 but is similar. I get excellent very sharp scans and can make adjustments before tweaking in PhotoShop. I scan with the emulsion side facing up. I use the film holders that came with the scanner and do one frame at a time so I can make the adjustments I desire. I'm not interested in volume so scanning a roll is not an issue. There are other film holders available and the suggestion of glass would work but I would get a thin flawless piece that would not introduce artifacts. You could also use a frame (sprocket) image after scanning just in case you did have to crop or put the 35mm film in a 120 holder.

HOME PAGE

 

[holmburgers]

Thanks guys.

David, you do make it sound like a good option, but I just can't imagine taking 36 pictures with a macro setup being faster than scanning. I don't know... maybe you're just damn fast!

Jeffrey, I think that I would probably be having no complaints if I was only scanning 120 & 4x5", but in general, the 35mm scanning just isn't popping. It's not bad, and certainly adequate for posting online I s'pose.

Did I mention I hate dust?

 

[dugrant153]

Thanks for posting this topic. I've been doing a comparison of prints between medium format and 35mm... and I'm finding my 35mm prints (even the really good sharp ones) to be soft as compared to 120 film. I think the Epson 4490 (which you and I both have) and other flatbed scanners are great for 120 film and up... but 35mm film seems to lose resolution. Enough that I personally notice on my prints. *sigh*. Oh well! Shoot 120 film for me!

 

[David A. Goldfarb]

David, you do make it sound like a good option, but I just can't imagine taking 36 pictures with a macro setup being faster than scanning. I don't know... maybe you're just damn fast!

Just to clarify, I'm suggesting two possible options--shooting a proof sheet with the whole roll under glass on a light box, which is just one shot, like a contact sheet, and shooting 36 individual frames. I don't see much reason to shoot 36 frames, just as I wouldn't scan an entire roll of film typically or print an entire roll in the darkroom, but if you work that way, I think it's much faster to take 36 essentially identical shots than to scan 36 frames with a scanner.

Most copy work involves some effort to set up the first shot, but then as long as you're shooting objects of similar size, or exactly the same size in this case, it's very quick. If you set things up carefully, you often don't even need to refocus from frame to frame--just line the neg or slide up on the LCD screen or a tethered computer monitor with live view.

Desktop scanners of the sort that most of us are likely to use are slow. How long does it take to capture an image on your film scanner at its native resolution once you push the scan button (typically from 20 sec. to a few minutes)? Now how long does it take to capture a photograph with a DSLR when you release the shutter (virtually instantaneous, and depending on the camera, you can often shoot directly to your hard drive via USB without having to transfer from a card)? Multiply by 36, and that's quite a bit of time saved.

If there are settings you like to adjust in the scanner interface, many of those are post-processing anyway, and you could do them as batch operations in Photoshop. You've also got total control over exposure with a DSLR to get the most out of under- or over-exposed images that can be hard to scan.

What you don't get with a DSLR is Digital ICE, so if you depend on that for dust reduction, you might be better off with a scanner that has that feature.

 

[michaelbsc]

...
Desktop scanners of the sort that most of us are likely to use are slow. How long does it take to capture an image on your film scanner at its native resolution once you push the scan button (typically from 20 sec. to a few minutes)? Now how long does it take to capture a photograph with a DSLR when you release the shutter (virtually instantaneous...

First let me qualify that all I have scanned to date are essentially electronic proof sheets.

But as for the slowness of the scanner, I just start it and walk away. I suppose the work flow will be very different when I try scan for a digital negative.

 

[glhs116]

I would only add that it is a mistake to think that the scanner doesn't give you control over exposure. It does. And it does so on a per channel basis which is pretty important for negatives. On a DSLR all colours are exposed at the same level. White balancing the RAW file is only boosting a blue channel that -- without correction -- is two to three stops weaker exposed than the red channel (assuming you shot without clipping red). Maybe for slides this isn't such a concern. When I was experimenting with DSLR scanning I actually went to the trouble of building up a pack of colour correction filters to make the negative close to a grey balance in order to address this.

I think DSLR scanning has its place and I definitely believe it can be quicker given the right setup. However, you will get my Coolscan 9000 when you pry it out of my cold dead hands.

 

[Zygomorph]

Since I started, not so long ago, in a wet darkroom, I have tried to maintain a very pragmatic approach to scanning film. If you have all sorts of wildy varying exposures and color temperatures on a single roll of film, you can only expect so much from any sort of automated process. Professional photographers worry/worried so much about exposure and filters and so on at the shooting stage for this reason: it makes darkroom work faster. The more consistent your input, the more consistent your workflow and output can be.

That said, Vuescan is very excellent for doing quick and dirty scans in bulk. Getting it and your scanner to behave essentially like a 1-hour minilab is possible, but the learning curve is a bit steep. And, just like a 1-hour minilab, you won't necessarily be pleased with the results. Beyond that, a good scan and print just requires a lot of experience, work, and patience.

 

[holmburgers]

I'm sure you're right, and the scans that I'm most frustrated by are usually errors on my part in exposure. I do try to get the best in-camera shots I can, but it's still true that scanners don't have the latitude of most films. You usually have to sacrifice shadow for highlight it seems like.

But again you're right... I just need to keep scanning and get better with each batch.

 

[Zygomorph]

Unless you're using a veeeeerrrry cheap or veeeeeerrrry old AND cheap scanner, they should easily be able to accommodate the latitude of color negative films, which are by their nature low-contrast affairs. How cleanly it does this is another matter, and is simply a matter of how well-engineered it is = $$$. Anyway, if you look at the histogram of a straight-scanned piece of color negative film, you should see that it has very comfortable digital margins. (Pro photogs inspect histograms all the time on digital cameras, or so I've been led to believe.) If anything is clipping, read below.

Most scanners are capable of varying the CCD exposure time, which can affect whether your shadows or highlights get clipped. This is just like making a proper exposure on a camera. Unfortunately, most software does not allow for the manual adjustment of this parameter. Silverfast does a reasonably good job: it prefers to get the highlights as close as possible to 255 without clipping, which makes sense insofar as shadow areas will be then be as far as possible from the CCD's noise floor. Vuescan is a bit more conservative about this, and also gives you manual control over the exposure time.

Anyway, the point of this is that any piece of software for which this often-critical element is automated can very easily clip either end of your tonal range at the scanning stage, which no amount of Photoshop Skillzs can recover. Figure out how your software handles this parameter, and learn to adjust it accordingly.

Another place where your toes and shoulders may be getting nipped is in any automatic negative-to-positive conversion (NegaFix in Silverfast parlance). I never use it unless I'm doing quick and dirty bulk work, which is almost never. This process is necessarily one that involves per-channel contrast expansion, and if you don't know what's going on or how to control it can also easily destroy highlight and shadow information you never knew you had.

This is probably enough info for now. Good luck!

 

[pellicle]

Hi

I have an Epson 4990, but this line of thinking should work for a number of flatbed scanners with exceptions here and there.

No exposure compensation: If I have a dense overexposed negative, or a dense underexposed slide, there's nothing but "computational" adjustment of the exposure, leading to noise and artifacts. How hard would it be to take a longer exposure at the CCD or to increase the brightness of the illumination.
Cheers!

I'm uncertain that this is the case. I feel that I can see (and measure) changes in exposure on my 4990 (and on my 4870) by altering the position of the dark slider when I use the view which splits the histogram into RGB adjustments

you can also fudge this by covering the calibration area with strictly neutral density material

however I've found that the benefits are not worth it in some ways.

may I ask which film types you would like to do this with?

 

[holmburgers]

Hi pellicle (and thanks to the great response from Zygomorph),

Perhaps you're right, and I admit to saying this without really knowing. However, I suspected it because sure I can change the histogram & what not, but any gross change results in very noisy and lousy scans. This was particularly noticable on a really dense b&w positive + screen-plate (a la Autochrome). There was no peering through this density, and the scans looked terrible. If the CCD exposure was manual, I could just expose it for 20 seconds or whatever and get a decent scan.

But, are you saying that once you do a preview scan, and alter the histogram/curve/etc., the final scan will apply those settings by altering exposure time at the CCD?

The bulk of my film collection is C-41, but I've got a fair amount of E6 as well, plus b&w.

 

[pellicle]

Hi

Perhaps you're right, and I admit to saying this without really knowing. However, I suspected it because sure I can change the histogram & what not, but any gross change results in very noisy and lousy scans.

I'm lost a little here ... by any gross change do you mean in post scan processing?

This was particularly noticable on a really dense b&w positive + screen-plate (a la Autochrome). There was no peering through this density, and the scans looked terrible. If the CCD exposure was manual, I could just expose it for 20 seconds or whatever and get a decent scan.

again I'm not sure what you're saying but recall that you are capturing what the film captured. Film is not a linear capture device so if you've got badly exposed film then there is only so much you can rescue from it.

But, are you saying that once you do a preview scan, and alter the histogram/curve/etc., the final scan will apply those settings by altering exposure time at the CCD?

again I'm uncertain what you're saying, but what I have has nothing to do with the preview. You change only the sliders at the bottom of the histogram tool within the scanner driver. NOT in photoshop of course

The bulk of my film collection is C-41, but I've got a fair amount of E6 as well, plus b&w.

if you feel that B&W needs any scanner adjustment it will mean it is just unusable, black and white will fit neatly within the capture range of the Epson scanner

C-41 will allow a little leeway, but expect noisy skies due to the increased noise in the system. Recall that with negative you'll be reading through a very dense media and therefore the light falling on the sensor will be less. I have not achieved noise free skies on my Nikon scanner which does have significantly alterable gain control. It just is that when you get exposure 'wrong' that the linearity of each of the channels will cause you problems in getting decent colour matches from the neg.

I don't know if you read this post, it was written about my 3200. I have found that it holds true across the other Epsons I have

in my view ...: Epson 3200 more optimal scans

Note carefully that the positions of the curves for R G and B fall at different densities on the negative; blue for instance is dense as red where blue begins as red is half way through.

 

[Zygomorph]

But, are you saying that once you do a preview scan, and alter the histogram/curve/etc., the final scan will apply those settings by altering exposure time at the CCD?

Different software will calculate exposure time differently. Unfortunately, it seems most often the case that how exactly they do this is usually poorly documented. Vuescan seems to be the best in that regard.

The only advantage I can see in calculating an exposure time off of a preview image which you've tweaked is for transparencies whose dynamic range exceeds that of the scanner; or if your scanner is so noisy that important shadow areas (or highlights, for negative film) need to be placed as far as possible above the noise floor. That is to say, you would be deliberately sacrificing detail in either the highlight/shadow regions of your image at the exposure stage of scanning for the benefit of the the other. Whether or not any software actually does this, I am not sure. I do, however, recall reading something about how the analog gain controls in Silverfast (if your scanner supports this feature) would simulate upon the previewed image the effects it would have on the final scan.

I only scan C-41 (very low contrast, esp. the new Portras), and with both my cheap and expensive MicroTek scanners, I've never had problems with noise in the sky, so I just use the HDR/RAW features on my software, and do all tweaking in PS.

As for noisy dense shadows... that'll usually just be a fact of life for slides and b&w. There are a few techniques at the scanning stage that somewhat alleviate this problem: single-pass multiexposure and multi-pass multiexposure. Most scanners and software support the first. A few scanners and software support the second. The first averages several exposures from the CCD for each line of the scan, effectively cancelling out noise to a certain extent (about 1 bit is gained with 3 exposures, 2 bits with six, and so on).

The second method scans the image several times (usually twice) favoring the exposure time either toward the shadows or highlights in each one, and then combines the resulting images. This can also be done manually in Photoshop, and might be preferred, since I've heard that registration errors crop up in software automated versions. To do this manually, and efficiently, would require that you make a preview scan at a known exposure, and, knowing the "real-world" dynamic range of you scanner, calculate exposure times that would compress your histogram to within those limits. All this all reminds me of so-called "compansion" schemes used in analog tape recording.

After that, there are PS techniques that I won't go into here.

To a certain extent, with engineering problems of this nature, you can get quite far by simply throwing a lot of money at the problem (i.e. drum scanners). But I have absolutely no qualms with figuring out how to get as much as possible out of limited resources, so long as I have hard data and a scientific rationale.

 

[Zygomorph]

As an addendum to the post above... I don't think that one should deliberately clip ANYthing at the scan stage having to do with color (except maybe in a multipass scenario...?), as this has the same effect it does in a digital camera: as the different color channels clip at different times, distortions of color will occur, like rainbow halos around the sun and suchlike.

Though this could be a cool artsy effect.....