Scanning Software, Fact or Fictitious?

[ljsegil]

I'm all for simple, and a simple rephrasing of my original questions would be: Am I losing anything in ultimate scan quality (Epson V750, Nikon Coolscan 9000ED) simply using the proprietary software versus using the sophisticated software at the scanning stage? Is there really a bang for the bucks ponied up for the big boys of scanning software, or can it be done as well in Photoshop after a simpler scan acquisition process? Or, to rephrase once more, at the level of sophistication and control offered by these (very much non-drum) scanners, is there much art to be learned to scan well at the acquisition phase, or is it all software manipulations after the acquisition and the software used for those manipulations is probably best selected based upon the user's skill with any particular program, with the actual program used not much mattering?
As far as I can tell from all my efforts to date, how I mount the film on the particular scanner seems to have a much greater effect on scan quality than any software package tweaking that I attempt to do.
But it all may be because I am just a simpleton, unable to properly use the sophisticated scanning programs to their fullest capabilities, and manipulating in Photoshop to cover my deficiencies as a scanner driver.
Wish I knew the answer, as I began this latest rant I do like simple, so I'm hoping the answer is proprietary software (maybe Vuescan to obtain DNG files) followed by Photoshop, and no need to invest further in sophisticated scanning software which does/does not have an actual impact on the actual scan data coming out of the scanner.
Hoping for clarity,
LJS
Of course, if it were really clearly an easily answerable question I don't suppose this thread would have continued as long as it has, nor would it be following on the heels of similar threads posted in the past.
Scanningly yours,
LJS

[bob carnie]

I am not sure if most agree with the following but we have been using four different scanners here and for all devices we find that setting good detail in highlight and shadow is critical, we do not sharpen at the scanning stage but rather in PS.
For highlight detail depending upon end media we will use LAB numbers between 92 and 96 or RGB between 245,245, 245 and 251 ,251,251.
for shadow detail depending upon end media we will use LAB numbers between 4 and 8, or RGB between 5-10 .
Each material is capable of different end points and practice will be your guide, I like the shadow with detail and highlight with detail to be present and the above numbers will give that result.

Regarding sharpening once again different medias will allow more or less, and I like to do this on the L channel or the K channel. Therefore we do not try to make a final image at the scanning stage but rather give as much detail as possible at both ends so we can work further down the line, currently we are scanning at 16bit to allow more moves in the editing stage .

When scanning we will immediately open in PS to check numbers to see if we like the scan..
hope this helps.

[Gordon Moat]

Some scanning software can offer a RAW data scan, after which you can adjust in PhotoShop. There is sometimes an advantage with this, depending upon the scanner, though it does add time to post processing each scan. The advantages of good scanning software are the accurate preview, and the ability to get a near ready to print file; these save time in post processing. However, if the preview in the scanning software is way off, or if the adjustments in the scanning software are crude or limited, then the results could require heavy work in post processing. Good scanning software saves you time.

Ciao!

Gordon Moat Photography

[Ken Lee]

1) Does a given scanner "expose" its controls so that an external driver can change the way the scanner actually captures the information ? Or does the scanner perform a capture, the same way every time, and merely adjust the results down-stream, in the drivers ?

Who would know ? The scanner manufacturer. Each scanner (or family of scanners) is different in this regard.

2) Does the given scanner manufacturer *publish* the proprietary APIs (Application Programmer Interfaces) that allow software drivers to control the scanner ? Or do the APIs only allow down-stream manipulation, after the capture stage is over ?

Who would know ? The scanner manufacturer, and software engineers who provide scanning software, like VueScan and others.

3) Do the controls provided by 3rd-party software actually let you get *better* images ?

It all depends on what you mean by better: The ability to rescue images that are poorly exposed ? Better image fidelity at the expense of slower scan times ? Faster scans at the expense of image fidelity ?

Your best bet, is to contact someone in Technical Support, for each scanner in which you're interested.

That being said, I repeat the assertion that each "adjustment" introduces a distortion, and that the fewer adjustments we make, the more natural will be the image.

Hence, getting it right at the scanning stage, is better than trying to rescue it in Photoshop.

[Bruce Watson]

...getting it right at the scanning stage, is better than trying to rescue it in Photoshop.

GIGO still applies. Yes indeed.

[Peter De Smidt]

[B]1)
That being said, I repeat the assertion that each "adjustment" introduces a distortion, and that the fewer adjustments we make, the more natural will be the image.

Hence, getting it right at the scanning stage, is better than trying to rescue it in Photoshop.

If by "getting it right at the scanning stage", Ken means that some are advocating doing a low quality scan and fixing it in Photoshop, then that's a strawman argument, as no one here is advocating that. On the other hand, if Ken means using the software editing tools in a scanning program on a good capture is somehow better than doing the editing in Photoshop, then the conclusion doesn't follow from the premise, in other words even if we should minimize adjustments, it doesn't follow that these software adjustments are better performed in the scanning software. If manipulation in the scanning software isn't optimizing the hardware elements of the scan, then it's hard to see how doing purely software adjustments in the scanning software is any better than doing it in Photoshop. Yes, as others as pointed out, sometimes there's a time-saving element of working in scanning software, especially if the software has a good system for dealing with color negatives; and in the early days Photoshop didn't deal that well with high resolution files, e.g. apply a filter to a hi res file and you might have time to go to lunch before it finished, and so getting the file as close to what was needed for a specific output in the scan was desirous. Today, though, Photoshop deals well with hi-res files, and the preview functions alone are much better than in any scanning software that I've used.

Finally, If Ken means that tonal adjustments entail "distortions," and "distortions" are inherent bad, then that's just not plausible. A while ago, I spent considerable time digitalizing and Photoshoping my father's slides from the 1950's. Many were badly faded, which often included color shifts. These scans required considerable editing, both in the scanning software and in Photoshop, to look natural.

[Kirk Gittings]

If manipulation in the scanning software isn't optimizing the hardware elements of the scan, then it's hard to see how doing purely software adjustments in the scanning software is any better than doing it in Photoshop

This is the crux of the issue isn't it? A couple of years ago I tried, without success, to get a straight answer from Silverfast on this issue related to the consumer flatbeds. In one vague public statement they had implied that their software controlled the inner workings of the hardware as has been described by Tyler above, but thier unwillingness to deal with this question in depth made me wonder.

[Kuzano]

Having worked with both hardware and software for many years, I believe that some of the most successful software over the years has been a mix of barely adequate software design and implementation, combined with creative and ongoing marketing, and the amazing accidents that take place in the industry.

Case in point. When IBM was seeking an operating system for their new Personal Computer project, there was DOS, as we all know, but there was also a lesser known and quite superior system called CP/M. DOS, in fact, was an unauthorized (and poorly modified) clone of CP/M. IBM was leaning toward the better system....CP/M. The unfortunate event in this story is that Gary Kildall, who owned CP/M, missed the meeting with IBM when they came to town to work on the deal. Mr. Kildall was attending another meeting. As a result of that major mistake, CP/M and Kildall lost the deal. To shorten this story, here is a link to "How Bill Gates outmaneuvered Gary Kildall".

http://arnosoftwaredev.blogspot.com/...ered-gary.html

Kildall later went on, with associates, to create a shell program that improved MS-DOS, called DR-DOS and recaptured much of the lost money from the IBM/MS-DOS union.

Why am I telling this story?

Well, it occurs to me that the software industry has been rampant with success stories wrapped around software that is really not that good, but is heavily marketed, thereby creating it's success.

Is Silverfast really a much better interface and controller of the scanner than the OEM software... Who knows.

But it seems to me that this discussion is much the same as making the choice in camera to run the software in the camera and create jpegs, OR to shoot RAW and avoid as much camera software processing. Likewise, it makes more sense to use the scanner to emulate a RAW image, instead of imposing yet another level of software.

I'm contending that it makes more sense to bypass as many levels of software as possible and learn to use just one piece of software very well.

As it stands now, from Adobe alone we have an array of choices, starting with the Elements, full Photoshop changing annually, Bridge, and Light Room. If we throw in the "supposed" control at the level of scanning, and if we are manipulating the images in the camera shooting JPEGs, how do we ever get to a finished image.

Does anyone ever get a final usable, no more post processing, image from Silverfast?

This is not a specific blast at Silverfast. I simply use them to make the point on overprocessing images from capture to "final image". Sorry if that's unfair.

When I think about all the discussions in this and other posts, my mind snaps back to the CPM/MS-DOS analogy.

It's been almost 30 years and still the most widely used software programs are often the result of massive marketing and unfortunate accidents in the marketplace.

[Ken Lee]

Peter, you are certainly right. I'm not talking about the corrections we would make to rescue a faded or poorly exposed image. I should have said so. I'm thinking about how to get the most faithful scan of an already lovely "fine art" image. Hopefully, most of our shots are in that category.

If we start out with a fine specimen (the Mona Lisa for example), then in a general sense, it's downhill from there, each time we perform an adjustment. The same would be true if we were to make a series of internegatives, or copy negatives. Even with the best lenses and process, every step is a "destructive" or lossy step, however slight, compared to the original: We lose some of the original data, and introduce some noise that wasn't originally there.

That's why, for the most lifelike images, it's generally best to make the best exposure, and develop as correctly as possible. If the next step is a scan, then it's best to get that scan as close as possible to the final result. If further adjustments are required, then it's best that those should be done as sparingly as possible.

[Scott Schroeder]

I thought I'd throw out a comment here. I'm by no means proficient in the whole scanning process. I could even say it was part of the reason I left Large Format years ago and went all digital. Well now I'm back but instead of scanning film, I'm scanning plates. I haven't made prints from those scans but just scans for the web. I was using Silverfast on my epson 3200 and the shadows were horrible. No matter what settings I used or adjustments it just couldn't pull out the subtle details.

Realize that this is for scanning a plate so it's more like scanning a print than a negative. On the wet plate forum Kerik suggested I try the epson twain driver. Well it beat the crap out of silverfast! I'm not sure why, but I'm sure it has to do with how both packages handle reflective materials.