I have a huge scanning project that involves archiving hundreds of old fragile sepia toned and B&W photos and many hundred documents from an estate. Customer wants all the images scanned at 2400 dpi and all the letters, documents and manuals scanned and saved as PDFs. Manuals can not be dissasembled or unstapled.
My current computers aren't up to the task. Help me build a system. What is hot AMID motherboard right now? Need something with large RAM capacity. Do I want to use SATA hard drives?
Also need recommendation for larger scanner. Something wider and longer than standard 8 1/2" X 11 capacity.
Any recommendations or experiances welcome.
Thanks in advance HW
Your computer's motherboard/CPU/RAM isn't going to make much difference, IMHO, unless you need to do some sort of image processing on every file. Worry about the scanner, and get a hard drive big enough to hold all of the images, and get a halfway decent midrange box, and be done with it.
Ummm,,,,,, 2400 DPI is a fantastically huge resolution.
A 1-inch square image @2400px and 24-bit color is around 6.35 megs. A letter-sized image would have to be at least 93.5 times that, or up near 600 megs for one image.
At that size, you could store--
--1 image on a CD,
--7 images on a regular DVD, or 14 images on a dual-layer DVD.
--a 200 gig hard-drive would only be able to store about 300 of these images.
Can you even work with a file this big?
I tried to resize a couple images as png's and jpeg's up to 26400 x 20400 pixels, using Photoshop 7--and it crashed every time (its maximum limit is 30K x 30K pixels, and most image file formats don't even allow images that large anyway). Of course if some of these documents are B&W then scanning in greyscale saves a lot of filesize, but unless they intend to view these files greatly magnified, there's not much reason for the extreme detail.
They probably just looked at their scanner and saw that 2400 DPI was the maximum, and wanted it,,,, -but imaging at this resolution and size really approaches what is practical to store and retrieve on desktop hardware, using regular software--and there may be size limitations of PDF's anyway, I haven't even checked into that. A PDF pre-loads all the images....... Assuming it allows that much image data, are they going to want a PDF that takes 5-10+ minutes to open on a typical PC? Probably not.
I would suggest that you ask that they consider a smaller scan resolution. Standard commercial printing is done at 300-DPI and 8-bit color; anything they submitted to a print shop that was higher-detail than that would probably just get downsampled anyway.
I would suggest you get a sample photo and scan it at 300, 600 and 800 DPI, and have these printed (by a commercial printer) and show what the difference is(n't). If the source document is color, leave one set of prints at 24-bit color and flatten a second to 8-bit color, and show them. (label these on the image before printing, you probably won't be able to tell the upper ones apart) My bet is that 8-bit color (or greyscale) @ 600 DPI is going to be sufficient.
As far as scanning things without damaging them, there are archival scanners, also called book scanners. These typically cost up in the thousands of dollars. www.imageware.de/en/ is one company that makes them.
A possibly-cheaper alternative is if you have a higher-megapixel GOOD camera (with macro focus), you may be able to use that with proper lighting. But even then--a good camera will run near $1000 itself.
Listen to Floppy_833
Is there some way to compress the scanned image data? I assume so.
With you having zero experience on the subject, how did the customer find you?? By accepting the absurd specifications without realizing much about paper, printing, etc, you may have dug yourself a deep hole that could hurt you financially. And if you are bound by a contract, just because it is a bad deal does not make it an invalid contract. Good Luck.
eta: setup a photo stand with good lighting and a semi-fixed camera
eta2: see the imageware site -that is like a photo stand but with better tools for your application
The photographs they want as gray scale and will not be imbedded within a PDF. The PDFs are text only or mostly text and will not be scanned at 2400 dpi, probably 300 dpi. They want the photos at 2400 dpi anticipating future unknown technology. They are not interested in anything less. The access window to this material is relatively short. Opportunity to rescan at a later date will be impossible. This material is of a significant historical nature and I get the feeling they want to cover their butt as far as quality. 2400 dpi grayscale isn’t near as big as color and more than 7 files will fit on a DVD. They are paying for the materials, so it doesn’t matter anyway.
I quoted the job as a fixed price PLUS an hourly rate and they agreed. I can spend 5 or 6 grand on equipment and still make a rather nice sum. This is a really big job.
Some of the photos will be save unedited, while others they want enhanced by increasing darkness and contrast. Some of these photos are starting to fade. Some rather badly. Thus I can’t have computer crashing on me because of lack of horse power. I don’t need any fancy sound cards or ability to play the baddest game. I need something fast and reliable that can burn DVDs and scan at the same time.
Scanner, motherboard and hard drive recommendations is what I need. All this discussion about file size, etc is pointless. The file specs are already a done deal. I can’t change their requirements.
I have many years experiance, just not any with such large file sizes. I don't have a lot of experience with the current crop of motherboards. I have older HP and a Microtech scanners, but need one fast and of a larger format. Thus the request for suggestions.
Your thoughts and opinions about any financial gain or loss wasn't requested and as far as you are concerned, not a issue. Frankly its none of your business.
Nothing in your reply addressed any of my questions. It's only purpose appears to be a puny effort to belittle me for taking on this job. I could care less what YOU think about my business dealings.
For the larger pictures, you may want to look into a short-term lease for a sheet-fed wide format scanner from Contex or Océ (built by Contex). I used to support those scanners for Océ's help desk, and they're top shelf machines. They handle fragile originals very well, and can adjust for thickness for things like mat boards. The software that's available for them will allow you to scan in a batch, making the process a bit less labor-intensive for you... just click "go" on the software, and stand over the scanner feeding the originals in.
I don't believe those scanners are capable of 2400 DPI, and I seriously think that you should educate the customer about all the extra time and grief that would be involved in getting that... with no discernable return at anytime in the future. If they can't be reasoned with in that regard, I'd almost have to recommend walking away from the job. 800 DPI is more than they'll ever need now or at any forseeable point in the future.
I'd forget about trying to scan and burn from the same PC at the same time... if you try to scan something that big while you're burning, the PC will be tied up for much longer than the buffer on the DVD burner can hold out. This means that you'll never complete a burn, and just waste discs. This won't matter much if you use the batch scanning setup mentioned above... you can just burn everything after you're done scanning.
I'd suggest that if you're looking to upgrade your computer to handle this job, that you get a decent Nforce4 board, and a dual core AMD. An X2 3800 will cost you about $300, with a motherboard about $80-100 depending on which one you get. Definitely at least a Gb of memory, preferably 2.
Make sure you have a fast hard drive with lots of capacity.
Well I'm not one to argue with a paying customer, was just inquiring.
Anyway--I have been told that the image file formats are big enough:
For JPEG - 65535 x 65535 (JPEG Standard, Page 40, table B2).
For GIF - 65535 x 65535 (GIF 89a Standard, page 12, para 20c)
For PNG - 2147483647 x 2147483647 (PNG Specification, paras 7.1 and 11.2).
For BMP - 4294967295 x 4294967295 (Bitmap Format)
....Also someone with Photoshop CS2 says its normal limit is 300,000 x 300,000 pixels. So maybe my older version is just stuttering or something. Someone else noted that GIMP will allow up to 16,777,216 pixels square, which would result in a file size of 786,432 GB..... GIMP seems to support regular TWAIN drivers, last time I tried using it.
I would advise you to stick with PNG format just because it's fully-open. And you'll probably be delivering it to them on an external hard-drive, which will cost $125 or so to put together (buy a Seagate internal drive and an enclosure). For the business docs, scanning them at greyscale/300 dpi, sweeping the color curve and depressing to greyscale will definitely shed a lot of file size.
If I was doing this, (for hardware) I would--
1) get a faster dual-core AMD mobo, and max out the RAM.
2) Use at least five hard-drives: one for the OS, one for programs, one for the swap file and at least TWO more bigger drives (300+ gb) for actually saving the files onto, until you finally transfer them to the external (save your work onto BOTH big hard-drives). SATA isn't really faster than IDE, but it is newer and more mobos support it now.
3) I'd get 74Gb Raptors for the main three hard-drives and put the OS and boot-optical-drive on the mobo, and run the other hard-drives off of a controller card (that's what I have now :D ).
The CPU will be about $300-$400
The 2Gb RAM will be about $160
the mobo will be about $100
The triple raptors are about $160 each, or normal drives would be around $100 each
the two extra BIG drives might be $200 each. Alternately, you could just buy an empty enclosure and when you're done with the job, just stick one of the big internals into the external enclosure and then give them that. Or both HD's, if they feel like paying for two copies.
(HD's are actually a pretty good long-term storage medium, as long as they are not hooked up to a computer)
So.... it'd cost $1200-$1400+ to get this stuff, and re-use whatever parts you don't buy (like the videocard).
The other matter of the scanner--is it possible to scan all the materials on a flatbed scanner? If you looked on the ImageWare site, you see that most of their scanners only do 400-500-600 DPI. It's only really flatbed scanners that have the 1200-2400+ DPI resolutions, because they can use close-focus optics and are meant to be used for scanning film negatives.
I know of at least one scanner that could be used for books without disassembling them; it was a flatbed/LED scanner that was see-thru, and has a stand. But one edge scans very close to the edge--like 1/8" inch or something. I am having problems remembering it now. It was a lot more than typical consumer-flatbed scanners but it wasn't as much as the archival/overhead type scanners usually are ($3000-$5000+).
2400ppi is a HUGE resolution. close to what high-end image setters pring film at, but then it's at about a 300 linescreen
IMHO, that's only practical for pace-size documents if it's text or line drawings, and then reduced to a 2 bit (B&W only, no grays) image. i mostly save stuff for printing at a .tif with LZW compression.
now, if you're scanning film, say a 120mm square transparency and you're going to be blowing it up for a two page print ad or poster, then you'll scan at the high rez, but the actual ppi at the final size will probably be closer to 200. most HUGE prints like billboards don't need super high rez because of viewing distance. many billboard printers only image at 9ppi. (yes, nine)
for the platform to work on, get a Mac G5 tower, load it with RAM and a pair on nice fat SATA drives set up as a striped array. save files to a terabit NAS, and deliver them on DVD.
using lower rez versions of the images and iMovie, you can make a catalog disk that will let the client view the images on a TV, and then find the storage disk for the hi rez version.
i understand that the client WANTS 2,400ppi, but my 20 years of experieince tells me that it's not practical.
don't know what kind of manuals these are, but if they are hard or soft cover books, they can be VERY hard to scan because they don't want to lay flat on the scanner.
i've worked with HUGE stat (reprographic) cameras, drum scanners, flatbed scanners, etc. anything "bound" is a PITA to scan... especialy if the inside margin (towards the spine) is small.
if you want to preserve the sepiatone look of those images, you'll have to scan those as color... scanning sepias as grays just doesn't seem right from either an artistic or historical viewpoint. however, i've been known to make grayscales into sepias so it's not excatly "wrong".
for faded photos, a few quick tips; don't use the "brightness and contrast" controll in photoshop, play with "levels" and "curves" and multiple layers of the same image, with some layers set to "multiply" are also your frined. if the fading is uneven, use layer masks to counter the fade.
you missed this:
and from your original post post it is my observation that, although you seem to have experience with small things, you obviously are trying to take a huge step and need advice, etc from us...my observation was not intended to belittle you, sorry if it did....I was just pointing out what can happen - advice is still advice whether it paints a rosy picture or not...
just get a Dell XPS or their Current Workstation model with 2 gig rams and several hard drives. Think either can handle terabyte drive combos.
No brainer. Get a Power Mac G5.
Can be configured with up to 16GB of RAM.
Wow, did you accept the job and sign a contract already? You should do a few images and time how long it would take you to do this thing, and the HD space necessary to do it.