PDA

View Full Version : Shouldn't these drum scans be better than this?



paulr
8-Dec-2008, 13:53
I had 20 drum scans made from 2-1/4 inch negatives. This was based on the assumption that the scans I'd already made (and laboriously color corrected, burned/dodged etc.) with my Epson 4800 wouldn't be good enough for exhibition prints. I dreaded the project, since it meant doing all the work over again, but I wanted the prints to look as good as possible. To this end I ponied up and had the scans done by the lab with the best reputation in NYC.

Now that I'm working on the scans, I'm not at all sold on the quality. They're sharper than my unsharpened epson scans, but not the sharpened versions. And shockingly, they have worse dynamic range. I see more shadow detail in the epson scans than I do in the drum scans.

I also find the color in the drum scans to be exaggerated. The greens are almost fluorescent, and all the colors need to be desaturated somewhat (the film used was Fuji portrait film ... not velveeta or anything similar).

Does this sound odd? I've never had drum scans done before. I wasn't sure what to expect, but I really didn't expect this.

Ken Lee
8-Dec-2008, 13:59
Did you ask the people who did the scanning for you ?

They would probably be very interested in making sure you are happy with the results.

Ash
8-Dec-2008, 14:02
It may the format they're saving them.

What colour space? Are they uncompressed tiffs?

Kirk Gittings
8-Dec-2008, 14:07
I have found drum scanner operators who are great at transparencies and b&w film but are incompetent when scanning color negatives.

paulr
8-Dec-2008, 14:51
I haven't talked to them yet, primarily because my sales drives me nuts (he gave me a good deal, though, at least on paper).

Nothing funny about the file format. They're uncompressed tiffs, Adobe RGB.

QT Luong
8-Dec-2008, 15:06
I've no experience having negative film scanned, only transparencies. That said, I have many 35mm images scanned on both scanners, and found the difference between the LS4000 (a dedicated 35mm film scanner) and a Tango scan to be quite noticeable, both in terms of sharpness and detail in shadows.

Noeyedear
8-Dec-2008, 15:19
Are you sure they are drum scans? Some years ago a printer scanned some images for me on an expensive flatbed, I thought the colours looked child like, the greens were very green, reds very red etc. Does the exif say anything about the origin? I have a drum scanner and colour negs can be hit and miss for me. One thing I did notice recently when I scanned some B&W on my old broken Artixscan 1100 was that they had better DR than my drum scanner. I think if it's not a late model drum scan the DR could be less, on tranparency it does not matter so much, in fact my drum pulls detail out of shadows from transparencies that has me going back to the original to see if I can see it. If you could post some samples it would be good.
You probably put a lot more work into your scans than the drum operator and did you have more dpi on the drum scans, if you are looking further into the negs from the drum scan it might not look so good.

Kevin

Bruce Watson
8-Dec-2008, 15:26
Now that I'm working on the scans, I'm not at all sold on the quality. They're sharper than my unsharpened epson scans, but not the sharpened versions. And shockingly, they have worse dynamic range. I see more shadow detail in the epson scans than I do in the drum scans.

Just to be clear we are talking about detail from the least dense parts of the film, yes? Either way something is wrong. A good drum scan often shows more shadow detail with negatives than do consumer flatbeds. Surprising but true in my experience.

As for sharpening, you aren't comparing apples to apples. Sharpen both files and see what you think then. PMTs are sharper then CCDs - that's just the laws of physics. But drum scanners can be picky beasts; they want to be fully tuned up to do their best.


I also find the color in the drum scans to be exaggerated. The greens are almost fluorescent, and all the colors need to be desaturated somewhat (the film used was Fuji portrait film ... not velveeta or anything similar).

I do drum scanning for hire and specialize in LF and negatives (both B&W and color). I've scanned a lot of color negatives and never had a complaint like yours, but your post of 10-Nov-2008 about this same subject piqued my curiosity so I've done some investigating since then. I think it comes down to the huge working space that most drum scanners use internally. The working space is huge because the drum scanner has to be able to handle the most contrasty, dense, saturated film you are ever likely to throw at it.

In very simplified and general terms, what happens when you drum scan is that you mount the film, spin the drum up, define the location of the film on the drum for the scanner, then make a preview (low resolution scan). From the preview the software pulls a representative sampling of data for your film. From this it builds histograms and things for the operator to view and work with. When the operator sets his black and white points for each channel, the software sets the scanner's hardware to reflect the operators choices. Specifically, the analog limits for the log amps for each channel are set. Most low end scanners can't do this, and it has ramifications.

The advantage of this is that the scanner can use its full digital range to represent the density range of each channel on the film.

The disadvantages is the same. And that's what I suspect is happening in your case.

An example from my experience: I found a scene with a subject brightness range (SBR) of 1.0 stops. (http://www.achromaticarts.com/big_image.php?path=smokies&img_num=11) It was an early morning in spring and raining; extremely flat light. I captured in with 5x4 160PortraVC. When I drum scanned it "normally" by setting the black and white points for each channel normally, the resulting image file "stretched" the density range out to be a full range image, with black being zero and white being 256 (say 8 bits for the sake of argument). It stretched all the colors too showing me things I didn't even know where in the scene. And yes, it stretched out the saturation too. I had to dial it back a fair amount in Photoshop.

This type of stretching is usually seen just in color negatives. Trannies tend to fill the scanner's internal color space better so it's seldom noticed. B&W negatives are mostly about just the luminosity axis so this is seldom a problem outside of some intentionally high key images (think heavy fog). Color negatives, OTOH, tend to be lower contrast and lower saturation. And... most drum scanner operators see few negatives so have little experience with negatives.

That, I think, is why you are seeing this "exaggeration" as it were. It's not a bad thing. It's only marginally controllable with most drum scanner software, and it's easily corrected in a photo editor like Photoshop.

Bottom line, negatives aren't trannies. Because of the inverted color and the orange mask, the operator doesn't really know what the scene is supposed to look like. As a drum scanner operator, I always try to err on the side of giving more data in a scan rather than less. That's what I as a photographer want; most photographers want the same in my experience.

Stephen Best
8-Dec-2008, 16:04
Just getting a "drum scan" means nothing. I've seen some pretty lacklustre drum scans (from high end scanners) in my time, most of which cost the customer serious money. It's more about the dialog with the customer, knowing what they expect from the scan, what they're going to use it for, educating them on how to prepare the scan for print etc. The actual scanner used makes less of a difference than many here are prepared to admit. If you can't get the results you want, take your business elsewhere.

Nathan Potter
8-Dec-2008, 18:00
Paul, I know it would be an expensive pain but it might be worth sending a neg out to a very high quality custom scanning guy who you can talk to, say by phone. Describe your expectations and hopes. This would give you some sort of a benchmark. Somebody like Lenny Eiger comes to mind.

Nate Potter, Austin TX.

dwhistance
9-Dec-2008, 02:52
Nathan's suggestion is a good one as it will give you a base to work from as well as something to take back to discuss with your current scanner. I'd be inclined to contact Bruce Watson (2 posts above) as he specialises in colour negative film although I'm sure Lenny Eiger would also be a good choice. From their numerous posts it is clear that they both know their stuff when it comes to drum scanning.

David Whistance

Frank Petronio
9-Dec-2008, 05:25
You can't expect the scans to match "your vision" when somebody you never talked to is doing the scans. The scans might just require de-saturation and an auto curve adjustment -- and they'll end up being sharper and nicer afterwards. Most likely they're fine.

Like Bruce said, the scanner operator has no idea what to aim for when making the scans from color negatives. I still think color neg is the best film to shoot, but you do need to babysit it more than a chrome where scanner operator "matches the chrome".

Next time you order scans show them some of the reference files you've done on your flatbed and explain that they are fairly true to what you've seen, but you want to see the extended dynamic range and sharpness of the drum scan.

Comparing vendors and working with more custom, photographer-centric scanboys is probably the way to go. But yeah, a decent drum scan should blow away an Espon flatbed especially with 120 film. I stopped shooting 120 because there is no decent cheap way to scan it.

Lenny Eiger
9-Dec-2008, 10:38
I had 20 drum scans made from 2-1/4 inch negatives. To this end I ponied up and had the scans done by the lab with the best reputation in NYC.

They're sharper than my unsharpened epson scans, but not the sharpened versions. And shockingly, they have worse dynamic range. I see more shadow detail in the epson scans than I do in the drum scans.

I also find the color in the drum scans to be exaggerated. The greens are almost fluorescent, and all the colors need to be desaturated somewhat (the film used was Fuji portrait film ... not velveeta or anything similar).

There are a couple of things I might be able to add. A lot of good things have been said already. I will say definitely that the comment that Frank made about giving your scans to someone you haven't talked to is absolutely spot on.

Most scanners autofocus on the drum so if sharpness is an issue, this lab might need to send their scanner to be tuned, unless its a Tango we are talking about. The Tango is great at chromes, b&w negs, but is very limited when it comes to color negs. This isn't a sharpening issue, but an aperture/sample size one.

The lack of shadow detail is likely the mark of someone who set the endpoints haphazardly. A working drum scanner should be able to pull every bit of shadow detail that exists on the neg.

The color issue is another operator issue. Color negs are harder in that there is no reference from just looking at the neg for what the client wants it to look like. One can always "match" a chrome, but if there nothing nothing to match to in a color neg, then you get whatever the operator was thinking at the time. It may have nothing to do with you, or your vision. Best case scenario is to supply a print with the scan - even a bad one with instructions about how you'd like it to be different.

I had a scan in here last week from a client, who did supply a print. It was an early morning scene and I set the colors a little blue. I sent him a small jpeg of the first scan I did and we talked it over. I scanned it again twice until I got it right. I'm sure other top operators have similar stories, we all do what we can to get the results our clients' need...

I hope that helps...

Lenny

Paul Kierstead
9-Dec-2008, 11:10
First let me say I am very ignorant of drum scanners, but I am curious about one aspect of the operation, both technically and business wise.

I assume the follow: A drum scanner can have some parts of the analog process tweaked, like the analog gain in the receptor. For these adjustments, it is critical to have the operator get it right, since lost information cannot be recovered. Same for focus, obviously. However, I would *assume* that colour adjustments are digital, including balance and saturation. I also assume that the sensor does not put out more information then is stored, so the adjustments can be replicated anytime post process.

Given that, why would a photographer be concerned about the colour/saturation of the scans? In fact, every digital adjustment slightly degrades the information; if an operator adjusts it, and then you adjust it, you are losing over a single adjustment (admittedly not a lot when you do 16 bits/sample). Is it expected that drum scans provide the equivalent of a custom print: the result is exactly what is desired? Is this part of what constitutes the very high cost of drum scans? To me, that seems like a waste of money; what is worse, you'll pay the operator to correct, probably correct yourself again (creatively), and then quite possibly pay the printer to correct *again* if you are paying custom print prices.

To me, I think I would want my drum scans to contain as much information as possible (and note digital stretching of contrast,etc/ does not add information) with as little digital manipulation as possible. I'll admit most people have a bear of a time reversing a colour neg properly, so I might give a little there :)

Bruce Watson
9-Dec-2008, 11:29
To me, I think I would want my drum scans to contain as much information as possible (and note digital stretching of contrast,etc/ does not add information) with as little digital manipulation as possible. I'll admit most people have a bear of a time reversing a colour neg properly, so I might give a little there :)

That was my thought exactly. I call it a raw drum scan; that's pretty much the service I offer. I do delete the orange contrast mask from color negatives and invert both color and B&W negative scans -- all files go out as positives.

Lenny Eiger
9-Dec-2008, 11:35
First let me say I am very ignorant of drum scanners, but I am curious about one aspect of the operation, both technically and business wise.

I assume the follow: A drum scanner can have some parts of the analog process tweaked, like the analog gain in the receptor. For these adjustments, it is critical to have the operator get it right, since lost information cannot be recovered. Same for focus, obviously. However, I would *assume* that colour adjustments are digital, including balance and saturation. I also assume that the sensor does not put out more information then is stored, so the adjustments can be replicated anytime post process.


As in everything we do, there are layers. I am using a computer, do I actually know how the type is being displayed on the screen? OK, yeah, but what about at the lower level, and os on... The same is true of a drum scanner. I never adjust the analog gain - I have a program that is not unlike Photoshop with different, very specific tools that do things. The scanner recalibrates lower level operations on its own before each scan. I adjust the image to the way I want to see it using curves and in some cases a levels-type tool. It doesn't have the same degrading effect as levels does in Photoshop, as it is before the scan rather than after.



Given that, why would a photographer be concerned about the colour/saturation of the scans? In fact, every digital adjustment slightly degrades the information; if an operator adjusts it, and then you adjust it, you are losing over a single adjustment (admittedly not a lot when you do 16 bits/sample). Is it expected that drum scans provide the equivalent of a custom print: the result is exactly what is desired? Is this part of what constitutes the very high cost of drum scans? To me, that seems like a waste of money; what is worse, you'll pay the operator to correct, probably correct yourself again (creatively), and then quite possibly pay the printer to correct *again* if you are paying custom print prices.


This question has a few answers. (And some disagreement, of course.) First of all, when you look at the inversion of a negative without any corrections, for example, it is very far off of what you want. The red-magenta mask that is on the film inverts to a cyan-green colored mask in the positive. Getting from there to a great print can be very difficult for someone with some skills, possible a little less than a total expert. It's much easier to get a scan that is "in the ballpark". The colors are somewhat normal, more or less what you want and can be adjusted to exactly what you want with some adjustment layers.

In addition, scans are often flat, as it is easy to add contrast, but once it's added in a scan (or a Photoshop adjustment without a layer) it is impossible to get back to a softer image, complete with the detail you just tossed. Scans are made to give you every option of print you want to make rather than match exactly what you want.

I make adjustments, to get as close as I can to this goal. My software (DPL) loads the settings into the firmware of the scanner and the scanner adjusts its range to scan that way. This provides me with a perfect histogram of the file with the adjustments intact. This is different from most scan software that simply does a raw scan and then adjusts to settings after the fact. This is what I think happens, what the manufacturer tells me happens and the words can start a war on the scan list... It is why some people prefer to get a raw scan and do the corrections themselves. Regardless of which method is chosen, which software is used, it is clear that adjustments done at the scan step are less destructive than those done after the fact.

I hope that helps....

Lenny

Paul Kierstead
9-Dec-2008, 12:08
Bruce and Lenny, thank you for the information (and time!). It helps clarify things.

Edwin Beckenbach
9-Dec-2008, 16:00
I make adjustments, to get as close as I can to this goal. My software (DPL) loads the settings into the firmware of the scanner and the scanner adjusts its range to scan that way. This provides me with a perfect histogram of the file with the adjustments intact. This is different from most scan software that simply does a raw scan and then adjusts to settings after the fact. This is what I think happens, what the manufacturer tells me happens and the words can start a war on the scan list... It is why some people prefer to get a raw scan and do the corrections themselves. Regardless of which method is chosen, which software is used, it is clear that adjustments done at the scan step are less destructive than those done after the fact.

I hope that helps....

Lenny

Not unless you're using an unprofiled workflow and creating a new CMS file for each scan. All other adjustments are applied post-scan using DPL's color engine which is probably not any worse but also probably not significantly better than making the same adjustments in Photoshop.

Nathan Potter
9-Dec-2008, 16:02
I echo Pauls comment. I'm just getting into some digital process flow stuff and trying to get educated; so Lenny and Bruce great thanks for your insights.

Nate Potter, Austin TX.

Lenny Eiger
9-Dec-2008, 17:29
Not unless you're using an unprofiled workflow and creating a new CMS file for each scan. All other adjustments are applied post-scan using DPL's color engine which is probably not any worse but also probably not significantly better than making the same adjustments in Photoshop.

Maybe. For one, I do use a unprofiled workflow. All b&w gets custom generated CMS'es and some color, depending on how good the histograms are with a supplied CMS. If I use a supplied CMS, it's because the CMS is almost perfect and I only do minor adjustments. Then I am doing far less adjustment anyway (than from RAW scan to corrected scan). One wins either way.

Lenny

Edwin Beckenbach
9-Dec-2008, 19:47
Maybe. For one, I do use a unprofiled workflow. All b&w gets custom generated CMS'es and some color, depending on how good the histograms are with a supplied CMS. If I use a supplied CMS, it's because the CMS is almost perfect and I only do minor adjustments. Then I am doing far less adjustment anyway (than from RAW scan to corrected scan). One wins either way.

Lenny

Sorry if I was confused but you stated above that you never adjust the analog gain. My understanding is that this is exactly what the curve in the CMS file does. Nonetheless, aside from manipulation of the CMS curve nothing else is applied in the scanner.

Lenny Eiger
10-Dec-2008, 10:48
Sorry if I was confused but you stated above that you never adjust the analog gain. My understanding is that this is exactly what the curve in the CMS file does. Nonetheless, aside from manipulation of the CMS curve nothing else is applied in the scanner.

Let me be clear. What I meant was that I don't have a tool, slider, pulldown menu that uses the words "analog gain" anywhere in the program. I can neither confirm nor deny whether a CMS adjusts analog gain. My thought process revolves around imaging vs electronics, whenever I get into the electronics part of it I find that there is a plethora of missing information and I just get frustrated. I look at what happens to the images as they are scanned with different methodologies and that is how I form my conclusions.

It has been explained to me that the CMS is a compilation of a few things, not just the curve. However, it is pretty clear that there is some hardware register inside the scanner that gets this information and if they can apply a curve as it scans then this is a benefit over doing it after the fact. We apparently both agree that they can do this.

Lenny

IanMazursky
10-Dec-2008, 23:02
In NYC, there are few labs left that still know how to scan.
A fare chunk are using Tango's, Hell scanners and Imacon's (not a drum scanner).
Many sell Imacon and flat bed scans as drum scans. A poor practice but i have personally seen it and been a victim (years back)
I feel its a big problem in the industry but something thats not going to change.

But i see the biggest problem as the knowledge. A lot of technicians are not versed in drums scanning. Its as much an art as it is science.
They are thrown into the job head first or trained by someone who didn't know how to use it in the first place.
Some buy a drum scanner of the internet or through a broker thinking they can make money at it.
Someone in the company said they know how to use it but dont. They end up with very angry customers.
I end up with a lot of customers that way.
Service bureaus used to be king in all things pre press. Now, every lab and photographer has a scanner and a printer.
I have great respect for those who could run a Hell 3010 and burn plates. Ive always wanted to try it!