Open

by damned truths

submit your photo


Hall of Fame
View past winners from this year

Please participate in Meta
and help us grow.

Take the 2-minute tour ×
Photography Stack Exchange is a question and answer site for professional, enthusiast and amateur photographers. It's 100% free, no registration required.

I'm looking for method to scan over 6 million images for a huge digital archive (conversion from all kinds of sizes to one dpi/resolution). Obviously the man hours involved make this a big of a big undertaking and I'm wondering if there is such a thing as an industrial scale automated scanner i.e you stack up a pile of photos on a machine and it scans them to disk until it's done, only needing to be reloaded from time to time.

Has anyone ever encountered such a thing?

UPDATE

To answer questions which have arisen: no paper will be used this is photo to file only. Resolution: 600dpi max, colour scans. Nothing too huge, I'm guessing there wouldnt be much more than $20-50k available for the system.

share|improve this question
1  
I don't understand where one would get the idea from this question that the images are in digital form already. –  mattdm Jan 22 '13 at 15:37
2  
No its "huge digital archive" that indicates that they are on disk. Realistically, this is impossible to do within his budget if @alex really means place each of 6 million photos on a scanner. The current budget numbers for large scale scanning is about ten cents per piece of paper. Thus $600K is a budget, not 20K. –  Pat Farrell Jan 22 '13 at 20:40
3  
It's "for a huge digital archive". Not from. I'm sorry, but I really can't see how that's ambiguous. –  mattdm Jan 22 '13 at 23:38
1  
Wow, right @matt, I misread it. I still thing this is impossible. Needs clarification from alex as to why he expects this to work. –  Pat Farrell Jan 23 '13 at 1:18
1  
@Pat That all looks very grumpy of me in retrospect. I dunno, having a bad day over here or something. Sorry! –  mattdm Jan 23 '13 at 1:28

3 Answers 3

up vote 8 down vote accepted

There are a number of important details missing from your question:

  • What resolution do you require?

  • Color or black and white?

  • Does the scanner have to be able to handle a mix of sizes simultaneously, or can you sort them ahead of time so that all the photos in a given stack are the same? (Sheet feeders typically work best when the sheets are about the same size.)

There are lots of high capacity, automatic feed scanners out there capable of B&W or color scanning at up to 600dpi and 75, 100, or even 120 pages/minute. Here's one, for example. If your project fits within the parameters of what they can do, it doesn't matter much whether they're called "document scanners" or "photo scanners".

If you're planning to scan 6 million photos, you're going to want to catalog them somehow. If the photos are currently in some kind of order, it may be enough to preserve the order when you scan them and give each one a serial number so that you can find it again later. On the other hand, if there's other metadata involved, you're going to have to figure out some way to attach that data to the right photo. Figuring out an efficient workflow that minimizes mistakes and lets you scan enough photos so that you can finish in a reasonable time can be difficult. (I once worked on a project that involved scanning tens of thousands of pages, and the workflow problem was probably more difficult than any technical problem.)

share|improve this answer
3  
That one, which is only $32,000, is rated for up to 120 pages/minute. That does not mean it will in fact scan 120 real pages per minute, or 240 in two minutes or 360 in three. In theory, it could do 7200 per hour. Never going to happen. But if it could, it would still take 833 hours. This is close to half a year at 8 hours a day. @caleb is right, workflow, paper management, and all sorts of other things are much bigger than scanning. –  Pat Farrell Jan 22 '13 at 3:54
    
There are much cheaper (say, $7000) machines with somewhat lower scanning rates (60 or 75ppm). And scanning at lower resolutions will usually allow faster rates. You can always scale up the scanning rate by adding more scanners. Then again, you have to have somewhere to put all that data, a way to back it up, and some means of finding the one you want later. –  Caleb Jan 22 '13 at 4:25
3  
I would be very careful with scanning photographs in regular document scanners. I have a document scanner and it is perfectly fine for documents - Fujitsu ScanSnap s1500, however a few people have mentioned marks on photos that they ran through the scanner. Given how it works, I can easily imagine this happening (I haven't tried photos, but thin paper only). Unless a scanner is meant for photo scanning, I would be careful about using it for the task (unless the print is disposable - or you use a transparent carrier sheet). –  DetlevCM Jan 22 '13 at 15:12
    
@DetlevCM This is a very important point! When I scanned some photographs at home the result was not overly satisfying - I thing it's because some surface structure which is supposed to create the photo-ish surface effect. #Alex I suggest before buying anything worth multiple digits, ask the seller to show you how a sample batch performs, with respect to both time and quality on photographs. –  Tobias Kienzler Jan 23 '13 at 7:18
1  
@TobiasKienzler You may be reading too much into the question. I don't think the OP has said anything about a schedule. He or she may have a month or a year or longer to finish the task, and the effort and equipment will obviously need to be scaled to match whatever the requirement happens to be. A bigger question is budget: how many high speed, high capacity scanners will you wear out in the process of scanning 6 million documents, how much does it cost to operate and maintain them, and is $20-50K enough to cover that? The OP will have to do that math him/herself. –  Caleb Jan 24 '13 at 8:33

Let me answer your direct question first:

The document imaging folks have been doing this for ages. They typically are used for huge image scanning projects, like the US Census form processing. I've never seen one for photos, but that is really just a difference in pixel density -- everything else would be the same.

These are insanely expensive to buy, and are a royal pain in the rear to keep running. The big problem is paper handling. Paper tends to stick together when its too wet, or too dry, or too wrinkled.

Your idea of loading a big stack in a hopper, pressing a button and coming back in an hour to see them all processed is a dream. Never happens. Paper jams are a fact of life.

But you seem to be suggesting that the starting point is digital, not paper. Why in the world would you want to have paper in the middle? Just get some software to convert from whatever format you have to whatever you want. The software is simple and straightforward. The paper handling is a really big deal.

Also, the commercial scanners like we used for the Census are really expensive, as in six and seven figure to purchase one. Then add trained operators, maintenance, etc.

share|improve this answer
    
+1 - You should put more emphasis on the insanity of printing out 6 MILLION(!) images just to rescan them though –  Tobias Kienzler Jan 22 '13 at 7:45
2  
?? They are photos he needs to scan, no need to print them out, they are already printed. –  MikeW Jan 22 '13 at 20:45
    
Now that we've edited the question, I understand better. But I still doubt that the scope of this is reasonable. 6 million photos is going to take years to scan. –  Pat Farrell Jan 23 '13 at 1:20

Wait, are you saying your images are already digital but only in the wrong resolution? In that case you should really skip the printing-to-rescan part and directly use e.g. ImageMagick's convert which can easily batch-convert all images to your desired resolution.

Assuming you're using Linux and all images reside in a directory original (including subdirectories), do something like this to convert everything to 300 dpi:

cd original
mkdir ../converted
find -type f -exec sh -c '
  convert -units PixelsPerInch $0 -density 300 ../converted/$0
' {} ';'

(That PixelsPerInch is required according to this unix.SE answer, the find -exec sh trick is due to this answer and also manages spaces in filenames) Test it on a small subset first to see whether it works the intended way, though.

Anyway, this will save you about 99.9999% of the time, money, paper and brains wasted otherwise ;-)

share|improve this answer
    
I expect that the OP has images in various formats, but this approach is what I was saying, even if you have to write custom conversion software, doing it as digits is far better than any scanning plan. My estimate is that your 99.999% faster is conservative and it will be far faster than even that. –  Pat Farrell Jan 22 '13 at 16:48
    
The OP says nothing about "images in various formats", just photos in all kinds of sizes. –  mattdm Jan 22 '13 at 17:35
    
@mattdm I read the question, then Pat's answer and my subconsciouses remembered only the "huge digital archive" part in the question instead of the "for a huge digital archive" - that's why after posting this I re-read the question and asked for clarification. I'll leave the answer here just in case someone else happens to actually try what Pat and I first thought the intention was –  Tobias Kienzler Jan 23 '13 at 7:21

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.