IMO, take the best images, use them for a book on the DL&W, and devote the proceeds to care for the collection.

For scanning, I can see three possible objectives / outcomes:

1) The raw scan.
2) The raw scan processed to match the original.
3) The raw scan processed to improve the original.


If I were a collector of images or a research historian purchasing images from this collection, I would want item #2. As a second choice, I would want item #1.

Which one(s) of these three options will be offered?

There is no pat answer to that question, John. It depends on the particular material to be scanned. You want to aim for a resolution and setup that somewhat exceeds the natural resolution of the material. On one hand you want to capture it all, but on the other hand you don't want to waste megabytes capturing amorphous blobs.

You really need to do your own scanning, or at least start into large jobs before you farm out. that way you encounter the problems you'll hit, and sort them out, adjust and learn as you work. Then you can have an intelligent conversation with the guy doing the rest.

If you are not getting satisfactory results, there are 2 ways to solve it. #1 change physical aspects about how the scanning is being done. #2 Post-process the image using Photoshop etc. A scanner has a bunch of physical electrical "hardware" that generates digital data, typically color levels in red, green, blue. On top of that, your scanner software has a bunch of "virtual knobs" which you can tweak to change the resulting scan. These knobs don't really change the physicality of how the scanner scans, they are simply post-processing the hard data from the scanner.

For instance, I was scanning newsletters and I got a lot of print-through from the ink on the back side of the page. So I put a black sheet of paper behind the page I was scanning. This didn't stop the print-through, but did make it uniform so it would not interfere with OCR. That's an example of changing a physical aspect of the scan. Another, if you are scanning open-lid, is throwing a dark colored towel across the scanner.

When you are doing archival scanning, you should archive the raw digital data coming directly off the scanner hardware, without any post-processing. Here is why. (just for example, let's pretend there are 8 brightness levels, 0-7, actually there are hundreds.) The levels are

0 1 2 3 4 5 6 7

You think it's too dark, so you oh-so-cleverly turn up contrast by 10% in the scanner. It would be

0 1.1 2.2 3.3 4.4 5.5 6.6 7.7

except shades can't be decimal, so it averages to integer, capping out at the maximum value.

0 1 2 3 4 6 7 7

See, there's a big gap between 4 and 6, and that'll create a visible artifact. If that's what you archive, you'll be hating life later, when you adjust it again, 10% more.

0 1 2 3 4 7 7 7

This is terrible. The adjustments are actually interfering with each other. The first was made at scan time, and the second when I went to use the image. It's like I've lost 2 generations of image deterioration, if anyone remembers "generations" of videotape. If only we hadn't made any adjustments at scan time!

Now suppose we captured the raw data, and make only one adjustment, of +20% brightness, at the very end. We get 0 1.2 2.4 3.6 4.8 6 7.2 8.4. Rounded:

0 1 2 4 5 6 7 7

See, that's a lot smoother. There's still degradation, but only 1 generation instead of 2.

I think that is probably the best way to approach it. The scanning that I have had done commercially has proved to be the most frustrating process imaginable. It is like being in quicksand.

I make art paintings and have them scanned for making prints. Fine art scanning services aim at matching the original. But the match has to be dedicated to a specific printer, so it should be proofed to the printer that will be used for making the prints. In my case, the prints will be made by the people doing the scanning.

So, they do the scan, and then set their printer to run a proof print from the scan that matches my original according to their eyes. Then I have to go and look at it and decide whether it matches according to my eyes. If it does, we are done. If it does not, then they make a second proof, and I go back to see if I think it matches my original. The process might take 3-4 rounds of revisions before it is good enough. The whole process might cost $300. It is not possible for the proof to ever match the original to the point where the eye cannot tell the difference. It is theoretically possible, but it will never happen in practice.