Scanning a Magazine

January 24, 2005 9:48 PM

I want to scan an old Life magazine article and post it on the Web. The magazine is bigger than the scanner. How should I go about this for a.) best legibility of text, and b.) smallest file size? Any particular combination of dpi and scaling that I should try?

And, what's the best way to post 'em? Just thumbnails linking to JPGs of the article chunks and photos? (Never done this kind of thing before.)

posted by Vidiot to Computers & Internet (14 answers total)

Response by poster: Both the photos and the article text. Basically, I thought it was an interesting article (a profile of Johnny Carson from 1970), and I thought others might be interested in it, especially in light of his death.
posted by Vidiot at 10:38 PM on January 24, 2005

To simultaneously achieve the maximum legibility and smallest file size I'd first transcribe the article text to HTML.

If there are photos on the pages that are important and no individual photo was too large for my scanner I'd scan each photo at maximum resolution. I'd use Photoshop to clean up the scans and make my master photos. Using the master photos I'd next produce smaller versions by reducing the pixel dimensions of the photos to sizes appropriate for the HTML page layout. Finally, I'd fiddle with GIF or JPEG compression settings in an effort to meet my space constraint. If the space constraint is really tight one might have to settle for small photos. I'd try to place these photos on the HTML page approximately where they appeared in the original article, relative to the text.

If the appearance of the original pages is important to the character and style of the article and the individual pages are too large for my scanner I'd scan each entire page multiple times in overlapping sections at maximum resolution. Then, using Photoshop, I'd assemble a single image for the page using layers - distorting or scaling each layer, if necessary, to remove scan distortions. Next I'd clean up the master pages, paying particular attention to the seams. Finally, I'd produce two smaller images of each page, a thumbnail and a reasonable sized version. I'd probably make the thumbnails anywhere from 100 to 300 pixels tall, depending on the HTML layout. If space was a concern I wouldn't try too hard to make the text in the reasonable sized image really legible, since readers will actually be reading the HTML text - for example, a height of 400 pixels or a width of 500 pixels might be OK. I'd place each thumbnail, linked to the larger image, on the HTML page approximately where the original page breaks occurred in the text.
posted by RichardP at 12:23 AM on January 25, 2005

Just scan the page as many times as neccesary in overlapping sections and piece it together in photoshop.
posted by Edible Energy at 12:26 AM on January 25, 2005

And in terms of legibility and small file size, scan it at highest quality possible, then resize it and cut the file size in PhotoShop or whatever photo editor your using...
posted by Edible Energy at 12:32 AM on January 25, 2005

Sure, you can go the partial-scan route, then piece it back together. But if you have a digital camera, I suggest another way: lay the the article flat, then take a picture of it. Crop and you're done. A heck of a lot faster.

First thing I'd do is select the areas of text, then increase the contrast for legibility. No need saving all those shades of grey when it's printed on white paper. Leave the pictures alone, naturally. Anything destined for monitor output should be at 72dpi. The height/width is whatever size still allows you to read it. JPG Quality of around 7 with a height of around 1200 pixels should do you good, if I recall my LIFE text sizes well enough. That'll give you readable images around 120K in size.
posted by Civil_Disobedient at 12:43 AM on January 25, 2005

When you contact the publishers to get permission to post the article, why not ask them if they have transfered their archives to electronic media, and if they would be willing to give or sell you a copy?
posted by Deepspace at 1:57 AM on January 25, 2005

Scan it in landscape mode, then rotate and do OCR. After getting permission from Time-Life, of course.
posted by Kirth Gerson at 3:25 AM on January 25, 2005

If you do try to use a digital camera to "scan" the page, a few things to consider. Make sure your camera resolution is high enough to produce a usable image. Use a tripod, make sure you are parallel with the image. Check your zoom setting, too much zoom or wide angle will distort the edge of your image also don't use the on board flash. Setup light sources to the side to produce an even wash. Or your page will likely end up bright in the middle, dimming towards the edges. Also if it's shiny magazine pages you will get a bounce flare.
posted by MrBobaFett at 7:15 AM on January 25, 2005

If it's all black-and-white, go to a copy shop with a big, high-quality copy machine and use said machine to reduce the pages down to something your scanner can handle.
posted by ROU_Xenophobe at 8:21 AM on January 25, 2005

Seconding Xenophobe's suggestion. You'll save yourself a major headache. Use the color copier even with the reduce setting. Of course, it's a glorified scanner, but stitching software has never been up to my standards.
posted by ontic at 8:31 AM on January 25, 2005

Most copy shops won't let you copy copyrighted material, unless it's for educational purposes. Libraries are also helpful locations for obtaining scanner sized printouts from old periodicals. University libraries in particular can often obtain nearly any format of anything published, if they don't already have a copy.
posted by spaghetti at 10:37 AM on January 25, 2005

To rephrase part of what spaghetti said: you do realize that what you want to do is a copyright violation?
posted by WestCoaster at 1:12 PM on January 25, 2005

Response by poster: I do, and I've asked Time-Life for permission. Haven't heard back from them yet.
posted by Vidiot at 5:29 PM on January 25, 2005

Response by poster: $2000/year for electronic publishing rights. No web posting for me.
posted by Vidiot at 11:03 PM on February 3, 2005