Making Digital Adjustments Pre- or Post Scan.

Thanks for the advice regarding medium for large format scanned images. As you can see, I'm exploring the whole digital thang. I have another question:

After scanning a negative or transparency, one has the choice of either embedding or assigning the scanner profile when opening the scanned image in Photoshop. Usually, when generating the profile (using Monaco EZColor), one turns off all adjustments in the scanner's driver, things like adjusting histograms for brightness levels, curves for contrast control, contolling hues and saturations, auto-exposure level, etc. This means not employing any of these aids when doing an actual scan.

However, when making these types of adjustments, isn't it better to make them as far upstream as possible, versus in Photoshop? For example, it's better to adjust lighting deficiencies pre-exposure, versus post-exposure. Would the same be true for scanning? Is it better to make image adjustments using the driver aids pre-scan, versus making image adjustments post-scan in Photoshop? (I'm assuming that a correct monitor profile has been invoked.) For example, large adjustments to histograms to Photoshop levels can thin out the histogram and introduce gaps in the data. Would it not be better make these levels adjustments pre-scan? Do these pre-scan adjustments get made as the image is scanned, or added after the image is scanned? (If added after, one might as well do the adjustments post-scan in Photoshop.)

A related question. If one makes these pre-scan adjustments, can one still invoke the scanner's profile when opening the image in Photoshop? At some point, one should still adjust for the scanner's idiosyncrasies, shouldn't they? Put another way, how can one make pre-scan adjustments, and still adjust for the scanner's idiosyncrasies? (This assumes it's OK to make pre-scan adjustments, which is still an open question in this thread.)

To summarize my questions, is it better to make digital adjusts on film pre- or post scan? If the former, how does one still adjust for the scanner's idiosyncrasies?

Any adjustment you make in Photoshop will degrade the image somewhat. That's why it is recommended that you do levels, curves, color correction and other such adjustments on Adjustment Layers; this way, the initial image remains unchanged. When you bring your picture into Photoshop, it is best to have the levels and color info as close to what you desire as possible. What I do is to make as many adjustments in my scanning software as possible so that my initial image file, once in Photoshop, requires as little adjustment and manipulation as possible (usually some spotting, tweaks to curves, and any local manipulations).



The color management system you use should calibrate your monitor to a standard. There are also tools to calibrate your scanner and printer to a standard. If everything is calibrated, to a standard and/or each other, then what the scanner sees will be what is shown on your screen and then in your print (as closely as ink on paper can match a screen...).



I'm sure someone will be more verbose and detailed than I was.



-m

I agree with the idea of trying to scan the image so that it requires as little adjustment in Photoshop as possible. However, is it the case that the image adjustments (levels, curves, etc.) made with the scanner software degrade the image just as much as the same adjustments would in Photoshop?

The "degredation" is primarily quantization loss due to 8-bit storage of RGB values (do some histogram stretching and look at how many of the values are depopulated). Usually the adjustments in scanner software are taking place before the raw scanner data is coverted to 8-bit per channel data (raw scanner data is usually 12-16 bits per channel). Loss in PS can be reduced by storing original and intermediate steps in 16-bit mode, and by using adjustment layers rather than sequentially applying adjustments to 8-bit image data. The main focus of scanning should be to extract maximum usable data from the film in highlight and shadow regions.

The usual advice I've read is the opposite of that given above. The things I've read suggest doing as much as possible in Photoshop because photoshop's editing tools give you more options, are more sophisticated, and generally better than those found in typcial scanner software. I don't offhand know why it would matter from an image degradation standpoint whether an adjustment is made in the scanner or in Photoshop as long as both are in 16 bit. FWIW (probably not much) all I do with the scanner software is set the black and white points, everything else is done in Photoshop.

When using SilverFast for scanning, I usually set the black and white points (setting the white point slightly outside the white data showing in the histogram), and sometimes adjust curves, in the scanning software to come as close to the original as possible in the scan. All other adjustments (e.g. sharpening) are turned off in the scanning software. I often do a light touch of unsharp masking in Photoshop (again, to closely match the original), and then save that as a "master" file in .PSD format. Any other adjustments, I make in Photoshop (CS), usually in 16-bit mode, if possible. I then save post-adjustment versions to different file names, but also in .PSD format. In approaching things this way, I'm operating on the premise that the algorhythms in PS will generally be more sophisticated than those that could be crammed into a small driver or firmware program. But, I haven't actually tested that premise.

Hi Neil,

Please try this site, it has an amazing explanatory on colour and scanning management.


http://www.normankoren.com/

regards

My understanding is that profiling a scanner, any scanner, only works for chromes. The tools all assume chromes. There are no tools for negatives, color or B&W.

It's subjective anyway - even chromes are not WYSIWYG. Think about it - what are the odds that the scene you shot with your chrome film has lighting that is an exact match for the film's calibration? Only happens in the studio, and then only if you are trying quite hard to make it happen. Outside, all bets are off. Said another way, some color correction is going to be needed, profile or no profile.

And that's one reason there is considerable disagreement about the need, or usefulness, of profiling scanners. It's really a religious thing - either you believe or you don't.

As to when to make adjustments, this too seems to be a religious thing. Some people really want to do as little in photoshop as possible. Some want to do all of it in Photoshop. When Adobe was keeping Photoshop up to date, it was easy to argue for photoshop. Now, the updates come quicker in the scanner and particularly the RAW converter software. Perhaps the most up-to-date algorithms are with the scanner software now?

Basically, you pays your money and takes your chances.

Me, I'm scanning on an old drum scanner. No profile, since I only scan negatives (color or B&W). The scanner software hasn't had a major update (beyond bug fixes) since, what, 1998? Photoshop is considerably more up-to-date for me, so that's where I do most of my corrections, but not all. I still tweak the software settings for each image to capture exactly as much data as the image requires, no less, and not much more. All captures at 16 bits per channel of course.

Learning scanning is similar to learning to use a camera, me thinks. It doesn't take long to learn how to get close to the correct exposure, click the shutter, and develop the film. But it takes considerably longer to learn how to squeeze the last bits of performance out - to nail the exposure, nail the development, control the contrast index as required for the image. You'll find scanning is similar, maybe. Good scans come fairly easily. Excellent scans take more work and experience. So clearly, YMMV.

Two basic schools of thought on this issue.

1. Making all the adjustments with the scanner software.

This works well for a single purpose image file, but could potentially require the image be scanned a second time down the road for a different use. This is what I was taught to do, because at the time, the scanning software was operating in 48 bit mode, then saving in 24 bit mode. All the big tonal adjustments were therefore better made in the scanner, than in Photoshop, which was dealing with half the data.

2. Making the major adjustments in Photoshop.

This is a much more flexible approach, not only because if one saves an unedited version of the scan, they can always return to it and start fresh, but because with Photoshop's incredibly flexible editing tools, with fully adjustable layers and masks, it is not only much easier to do, but fully reversible in the beginning, middle and end of the editing process, as long as one saves an unflattened version of the file. The fact that Photoshop now is happy working with the full 48 bits of data the scanner exports, it is a no brainer.

That said, one can vastly improving their scanner results by simply setting the white and black points as mentioned earlier in this thread. Unfortunately, the quality of the software interface for the scanners on the market today varies widely. Most now include functioning histograms as a tool, but some are more accurate than others. I have a Nikon 8000 and an Epson 4870. The Nikon's software is very good IMO, with very accurate histograms and a wide selection of editing tools and options. The Epson software gets the job done, but the histogram is less detailed (and hence, accurate). I typically set my white and black points with a little breathing room, do a test scan at 600 dpi with a 4x5 transparency and open it in Photoshop to make sure I got all my highlight and shadow values. If I did, then I reset the resolution to the scanner's optical resolution and do a high res scan. That high res scan gets saved, opened, sized to 8x10 at 300 dpi, and saved as a discreet file for working on in Photoshop. I use the Layer Transfer Method which Rich Sieling at West Coast Imaging passed along to me, making all my adjustments to the 8x10 version of the file and then transfering all those layers to the high res file. Much quicker than working on full res images.