Scanning 8x10 Transparencies

[Lachlan 717]

The choke points will be mounting, dust removal, scan time, "write" time and demounting.

Unfortunately, I don't think that will be a quick way to do this... Detail=time.

[Peter Gomena]

Ditto. Detail = time. There's no fast way to do this given the parameters you've established and the equipment you're proposing to use.

The "choke point" is the scan time coupled with how quickly it can transfer image data to the computer and have it recorded on the hard drive. Firewire is better than USB-2, but only marginally so. There's not really a way to increase the rate of data transfer.

An 8x10" 8-bit color file at 1800ppi is about 750mb. It takes the machine a bit of time to generate and transmit that much data.

If you want to open the file and give it a quality check each time, opening the file is going to take a while, too, especially on a laptop. Add setup time, pre-scan time, adjustment time to the scan time, plus any data entry to track your work, and you will process about 30 original a day.

Peter Gomena

[Peter Gomena]

The best solution might be to ask the archives to do the work for you. It might be prohibitively expensive, but they may have full-time staff doing exactly what you propose to do. It will save you a tubful of woe.

Peter Gomena

[jccalvin]

Unfortunately the staff at NARA do not have the time or the equipment to digitize these transparencies for the public. I work with them frequently and it is a no go.

I have given up on the lap top approach. I am going to have to bite the bullet and cart a small tower PC and monitor up to the 3rd floor. Yes they do have elevators. But every piece of equipment (and pieces of paper for that matter) have to be screened and recorded by serial number going in and coming out.

Some of the factors slowing down the scanning can be improved. I can add a SATA-II 15K HD and that will help on the read-write. The Epson GX20000 has a SCSI-2 connector. I can insert a PCI Ultra Wide SCSI-3 card in the PC. There are cables to connect the scanner SCSI connector to the PC SCSI connector. How much improved throughput is hard to estimate without testing. If I can get up to 50 per day that is acceptable.

I do not have to be ready until September as my next project is later this month at the Library of Congress. This is relatively easy as I am photographing 14"x18" topographical maps. Scanning would yield better quality but I have to copy between two and three hundred maps a day. With a tripod mounted camera and a wireless remote I can whip through the maps and shooting Canon RAW the quality is acceptable.

[erie patsellis]

if you're shooting topo maps, my guess is you have a Betterlight back, if so there may be a better approach than a scanner for what you propose.

[Frank Petronio]

You could set up a darkroom and dupe them all onto new film in a day.

[Gordon Flodders]

I think that a tripod mounted DSLR (maybe a Canon 5DMK11 or similar) might do the trick, but problems may arise from barrel or pincushion distortion even with some of the better lenses.
I have not shot 8x10 transparencies but you could give it a try. Biggest time consumer is the setup, but once you have it right, you can go a whole lot faster than using a scanner. It may even be possible to pull the images into line in PS later? You'll also need a suitable light box.
It all depends on how large a final image you plan on making. Maybe a 100mm macro lens or similar might pull it off?
I don't envy your task.

GF.

[David McNiven]

Would stitching - say 4 shots - solve the resolution problem with using a camera? That would double the linear resolution wouldn't it?
Slower of course but nothing like as slow as scanning - and maybe transparencies illuminated for less time if that's relevant.
Hasselblad 38 Biogon was said to be good for this work IIRC.

[Peter Gomena]

Stitching maps is asking for trouble. There's just too much fine detail, and the details have to match perfectly at the edges - four times. Any lens distortion or tiny misalignment of the camera will give you grief. Been there.

The best way to copy original maps is with a roller-type scanner like those used for blueprints. Old, fragile maps or blueprints can be sandwiched between sheets of mylar and run through the scanner, giving 1:1 reproduction at 300dpi. Another good way is with a moving-bed scanning camera (about $100,000.) A Better Light scanback is good, too. The roller-type scanner is the cheapest and fastest option of the three. Since the OP is copying copy transparencies, he's stuck with scanning.

Peter Gomena