Workflow for digitizing photographs
I'm embarking a long journey: digitizing family photographs and slides.
After some nominal web surfing into the topic, I have to admit my head hurts.
I'm sure some of you are experienced with such a project. I am asking for tips and advice. I am seeking quality results as opposed to something "good enough" for web site thumbnails.
I'd like to hear what hardware and software you use, how you approach the work flow, etc. I'm not against proprietary software, especially considering some of the potential hardware involved, but naturally, I am interested in free/libre strategies and workflows as much as possible (not online or web based). Tips and links about post-processing too. For example, if I capture an entire album page, how to quickly cut and crop each photo to a separate file?
The family photograph collection consists of Kodak "Brownie" black and white photos from the 1950s (no negatives and the photos have white borders, sometimes with processing date stamps), Polaroid Colorpak photos from the 1960s and 1970s (no such thing as negatives :)), 35mm color and black and white negatives (preserved in glassine sleeves), and color slides.
Thousands of images.
At this time the only related hardware I own:
* Visioneer 3300 flatbed scanner
Maximum resolution of 600 x 1200 dpi
Connected to a Windows NT4 machine (connected to my network)
Only scanning software available is MS Photo Editor (actually works fine for scanning)
Not an appealing tool, but perhaps short-term "training wheels." :)
* Canon A590 IS point-and-shoot digital camera
* Camera tripods
For conversation sake, I am ignoring a budget. I'm prepared to spend money --- but I'm unlikely to spend thousands of dollars for Nikon scanners. :) Decent equipment and reasonable speed, yes.
Speed. So many images to convert. Will I live that long? How to work smart and efficiently?
BTW, I never used Photoshop or GIMP. Digikam has never been anything more than a database for me. :)
As always, thanks for your time and sharing! :)
I've never undertaken something like this before (and I get a bit dizzy when I think how I would start), but have you considered getting a professional to do it? I can't imagine thousands of pictures would be cheap to digitize, but this kind of project could be pretty time-consuming.
That's a huge workload to undertake on your own. I would have to agree with jprzybylski on that one.
But if you were to take on this animal the 1st thing you would need is something like these or this depending on your budget.
With a scanner like that you could line up 8-10 4x6's per scan and dump them on your HDD then worry about going through and cropping them later or cropping after you've scanned maybe 5 sheets or so. This is definitely going to take ages to do unless you have able bodies to help and y'all do it in shifts. Maybe you might wanna hire a professional instead. Or just pace yourself and do it little by little until one day there all done. I'm not sure what software would be the best to use but that shouldn't be to hard to find.
My approach is to experiment for a while before deciding on the equipment and process that will consume hundreds of my hours over the next few years. I'm only going to do this once. I'm not going to do it over ... I won't live that long.
I'm also converting hundreds of video tapes to digital. I'm finding it much easier to make decisions on the video project than deciding the best way to capture the photographs.
Time and Quality are two factors at the top of my list. I'll set up a MS Windows system without hesitation if it can do the capture job better or faster than a FOSS setup. What quality is good enough and how much extra money or time does it cost for each incremental increase in quality? All factors I'm starting to figure out.
I have a not-new Epson V500 flatbed scanner I acquired for some simple experimenting with prints, negatives and slides. A friend who has been into photography his whole long life isn't interested in this type of scanner. For his conversion project he'll be looking at a film scanner costing a few thousand dollars (but not the ones costing tens of thousands). His primary source objects are negatives not prints. I'm not recommending the Epson, just an example of my experimenting.
I have a corner of a room with furniture and equipment dedicated to this project. Once the workflow has been worked out, I believe the secret is to work on it a little bit each day having minimal setup/teardown time for each work session.
I think by taking the time to experiment and learn about equipment, software, quality details, and storage/database implementation, that I'll be making the most productive use of my time on this over the next few years.
Of course technology changes and equipment breakage may lead to changes in the setup I use.
Please let us know the details of what you learn along the way.
EDIT: Many years ago I found I got better copies of black and white images by copying them as color. You may want to consider capturing your black and white photos as color to see if there is any quality improvement. You can always remove any color artifacts later.
I recently did this with my Nikon DSLR and tripod (with retro position) to reproduce photos that are glued into an album. It takes some time though, but my computer is much faster with the photos than with scanned images. I scanned some photos that were not glued into an album and it took ages here, so that was no real choice for me, but I guess it depends on what expectations you have for a "perfect" result. I have yet to get a cheap light table to try it with slides (using a macro lense). If it will work, the resolution I get should be better thanwith most flatbed scanners.
A similar DSLR and macro lense should be around 700-800 bucks.
I did a few hundred, not a few thousand. I have an old HP 4300C scanner that was USB and worked with xsane/xscanimage. Its probably 10 years old so you could probably find something on a local craigslist for next to nothing.
Most of my pictures were taken with the same camera so all the same size. So I used painters tape and taped some cardboard to the scanner glass to line up the pictures in the same place every time. Then did one preview, adjusted the margins and would scan several pages at once just incrementing a number in the file name each time (Picture1, Picture2, Picture3, etc).
Next was ImageMajik to bulk convert to .jpg then flip through them in viewer looking for issues.
next step was exiftool to bulk set EXIF data. If I knew the date of one of the pictures, say it was written on the back or one of those cameras that put it in ugly yellow numbers on the front I'd set the exif all to that date as a start. Then a simple shell script that had several exiftool commands so I could run it to set the device digitized to the name/model of my scanner, date digitized, description, location, etc.
Once I had the base Exif set then I could edit one by one either with exiftool or a gui app such as kphotoalbum.
I set the scanner on a small rolling cart and would put it in front of the couch and sit the laptop beside and scan a batch while watching TV or something. I'd group batches such all my baby pictures in the first house, or wedding pictures or wedding reception pictures since they would all generally have the same base EXIF tags, i.e. same description, location, etc. places where I knew the location I'd look up the coordinates on a map and go ahead and embed those as I might remember where they were now but years from now might not so I wanted to fill in as much information as possibly.
I did this sort of thing a few years ago -- family photographs from the 19th and early 20th centuries -- with a H-P flatbed scanner (first a parallel then a USB, can't remember the models and the USB one is about 200 miles away). Most of those were from glass negatives and the prints were mounted on gray card stock "suitable for framing." That worked pretty well, but you sit there and feed the thing manually. There were (and probably still are) scanners that automagically load and scan from a feeder, you know, stack 'em and then feed, scan and save, the sort of thing you use to do OCR with. They do work, but you gotta sit there and watch. The H-P ones came with software that did all the clean up, color balancing and and the like for you and then you'd save the image file. And, of course, it was (and most likely still is) Windows software. Good time to think about VirtualBox.
I also did a few hundred slides with a gadget that you stacked the slides and let 'er run. Same idea, worked darn well, too. I just now searched for "slide scanner" with Google and it turns up a whole lot of equipment, much of it under $100 for doing that sort of thing.
A thought. There are photographic gadgets you can get (dirt cheap) for mounting slides? Back in the day, you did your own developing and mounting. You can get one of those and mount your negatives then feed them through a "slide scanner." Most of the software I've seen (and some that I've used) deals with both positive and negative images, might be worth a look-see (and it's not hard to cut and mount negatives with one of the gadgets).
This might be something where you just hold your nose and use Win7. You can scan with GIMP (and I think Image Magick) but it's a pain in butt when you're talking about large volumes if there's a piece of software that does most of the twiddling for you (you can always open a file in GIMP or Image Magick after the fact and tweak and crop to your heart's content). Personally, I find Image Magick easier to use, YMMV.
One piece of software that I did use with the H-P scanners was VueScan from Hamrick Software (http://www.hamrick.com/index2.html?u...SJfdalzohRMg.1). That baby works really, really well. I had one image from 1880 that was so washed out and had water damage that it was useless. I scanned it into VueScan, hit the "balance" button, and wow! Nothing else I tried worked worth a hoot. It also did a really good job with Polaroid, changing all the magenta and cyan to usable images that looked more like life. It's pay for, but I'll tell ya, you want that one (and it's Windows, Mac and Linux).
Hope this helps some.
Some really good advice already in this thread!
Think up a file naming scheme, and take advantage of xsane's filename counter, it will save your sanity! You will probably want to save the scans as tiff files, then enhance the tiffs, and finally export them as jpegs. Don't feel you have to keep the original tiffs forever. Be ruthless about file storage, thousands of tiffs will use a huge lot of disk space. Keep the prints and negatives, not the tiffs.
Images scanned from chemical film will always benefit from adjustment of contrast, white balance and saturation, and from sharpening. 90% of them will look really *great* on default auto enhancement settings, and then you can spend more time getting the other 10% right, straightening horizons, doing crops etc. The best part of scanning old photos is that you get a chance to rescue those underexposed disasters from thirty years ago but it's hard work ;-)
I would suggest using either imagemagick (or graphicsmagick), or darktable. Imagemagick will give you instant command line bulk scripting, but you will have to spend a few days reading the manual and experimenting to get the right parameters. Darktable is more friendly for seeing good results quickly, and you can apply the same saved preset parameters to more and more photos, then give each photo a final tweak before exporting them to jpeg (or flickr/facebook/whatever). If you use darktable, checkout the 'velvia' filter to make your scanned film colours more vivid. As another alternative there is Gimp's script-fu.
Don't forget the wise advice of Linus about backups - the best way is to let the whole planet mirror your stuff. So make sure all your family have their own copies of everything. Give them away at weddings :-) Everyone with a brain thinks DVD slideshows are lame, but it's a really good format for old folk who don't understand computers! See imagination and dvd-slideshow at SBo.
The other important thing about scanning old family photos is to spend time identifying people and places in the images. There is nothing worse than looking at eighty year old photos and not knowing who some people are or where they were taken. You can add this information in exif comments, or make some text files, or just print it on paper, but please make sure you record it somehow for future generations. Properly documented family archives are rare, and immensely valuable.
(sorry guys for the OT but)
welcome back David, long time no see! :)
Uh well sorry for the OT, but thanks! I'm starting to get back to 'normal' now that the archaeologists have finished with me :-)
... but (back on topic) the archaeologists have left me with a hard disk full of 400 Gb of RAW files and tiffs :-( ...
I have started this project also. My choice for a scanner is a HP Scanjet G4010. It was about $250 when I bought it. There is a driver for linux that works now, however, it does not work with the file/negative part of the flatbed.
My solution was to install V-box, and Win XP, and use the HP supplied scanner software to do the scanning. I can easily copy the scanned images via samba share.
Keep in mind the image you get will only be as good as the original. I have found that the colour on many of my slides ( some 25+ years old ) has changed in storage. I do find black and whites have stored much better. The images seem crisper.
This is an ongoing project for me. It is time consuming, and being retired helps.
One thing I would ask, if anyone knows of a good tutorial for editing old photos, using the Gimp, or any other application, please post.
Thanks for the comments thus far!
I'm not interested in hiring anybody. I'm too much of privacy advocate to trust my personal collection to a stranger, let alone holding my breath that the slides and negatives are not damaged, as well as the old photos, of which only the sole copy exists. Besides, I wouldn't learn anything. :)
Although a large project, I don't view the task overwhelming. That such a project is long-term does not discourage me. As the old adage goes, "Rome wasn't built in a day."
I need to learn some basics, to learn about appropriate hardware, as well as how to work efficiently and smartly. I'm intrigued by the many shoestring budget solutions I've seen, but I'm looking at many conversions, not just a few dozen. I want serious solutions and workflows.
My initial efforts into learning about this topic indicates that hardware is critical and that Linux plays little to no role in actually getting the images to a computer. After the images are on a computer there are plenty of software tools to help with post-processing.
I have read good reviews about Vuescan and a Linux version exists. That the software is proprietary is not a serious issue if the software helps with the task.
I use the OSE version of VirtualBox, which does not provide external USB support. I'm not convinced I need that support anyway, just to run Windows scanning software. I think as long as I find a quality SANE supported scanner I'll do fine with existing Linux software.
When I face large projects I look for ways to reduce them to smaller "bite-size" projects.
My first focus is a nominal collection of Kodak "Brownie" camera photos. The photos are on loan to me. I want to get them into my computer as soon as possible. Those photos are my first priority and require no special hardware. I can use my old scanner. I do need time and to learn some basics.
All of the photos are black and white, all are 3-1/2" x 3-1/2", with a white border. Originally I thought I'd use my digital camera to capture several photos at a time, but I don't think I can do that without suffering some kind of distortion to the photos at the outer edge of the collection arrangement.
Everything I've read thus far indicates that using a camera to capture old photos is a process of one photo at a time, whereas scanning captures multiple photos per pass. Further, all I have read thus far indicates scanning provides more than sufficient dpi resolution.
My existing flatbed scanner is slow, but I'm thinking a flatbed scanner is nonetheless the best tool for these photos. Because of the sizes, I can place several photos on the scanner in one pass. I can crop to individual files after they are on the computer.
Long term, a faster and improved scanner would help much, especially as I then can connect a faster USB scanner to my office machine and dispense with the older computer to which I have the current scanner attached.
After reading about this topic, I still don't know the optimal parameters for scanning. I keep reading that 300 dpi is sufficient for scanning photos (not negatives or slides!) as the resolution of the original photo is no better than 300 dpi. I've read that 200 dpi is sufficient for scanning photos.
Yet, should I scan as grayscale? Or scan as "color" and then use post-processing to restore each file back to "black and white"? I read one author's perspective that color is best to capture finer details (24-bit vs. 8-bit?) and then use post-processing to restore to black and white.
My primary concern with scanning photos is obtaining the best quality at the beginning. Post-processing helps, but I need quality images at the very beginning. Hence my focus on scanning parameters.
Scanning is slow, even with a "fast" scanner. There likely is no way to avoid that. :)
Regarding the negatives and slides, I've learned that getting negatives and slides onto a computer is a different game than photos. I've decided that a dedicated "film scanner" is necessary. I can take my time learning about that. Some of the hardware I've seen costs thousands of dollars, but I've seen hardware that costs "only" hundreds and requires no special software or unique operating system.
For now, I need to get those old "Brownie" photos onto my computer. :)
That said, there was a series of such articles about GIMP in Free Circle magazine. I never used Ubuntu but the magazine carries useful generic tutorials, of which GIMP and ImageMagick are topics. Look at issues 34 - 36 for photo touchups with GIMP. There were more GIMP tutorials in that magazine --- check the magazine archives.
I'll recommend VueScan as a pretty darned good front end for scanning (it supports something in excess of 2300 scanners) and has worked just fine for me with two H-P flatbeds. At $40 and it works and it does pretty much everything you want to do, well, beats heck out learning to deal with GIMP. Just as an aside, Photoshop and GIMP are really great tools for the folks that actually know how to use them and that knowing is a pretty steep learning curve. I just get frustrated by it and go back to VueScan or Image Magick and produce good results.
Advantage flatbed? It's glass, there's a cover, your photographic prints lie flat (and you can put a book on top of the cover to make the things lie flat if necessary). So do your negatives (the big ones). 35mm, you probably want a gadget (it doesn't have to be made of platinum, you know, it's just gotta work).
Advantage scanning in gray scale? You can set the resolution you want (and 300 dpi is more than enough for a good image unless you've blowing it up to 16x20 or something (and even then it's not bad). You'll probably find that scanning in color then converting to gray is... well, not so hot, better to just set the resolution to what you want. With B&W prints, gray scale is more subtle than color-covert, you get a better image at least in my experience.
If you've got a copy stand to mount a camera on with LED lights (so you don't cook everything), that works too. Actually, it works quite well but it's got to be at the right angle (or you'll get terrible keystoning; don't try to hand-hold or shoot vertically) and you'll get awfully tired of fiddling with it.
Flatbed for large jobs is a good way to go. For film and negatives, you really do want a gadget made to do that -- they backlight like a projector or enlarger does and give you a pretty darned good image. The newer ones have LED's which is a tremendous advantage because LED's are the same color temperature as sunlight, no color shifts one way or the other. They're under $100 and there are quite a few good ones to choose from.
H-P and Epson are at least two of the better makers of equipment and they're typically recognized by SANE without additional software (also by VueScan which is also image manipulations and editing software, kinda like Image Magick).
Anyway, hope this helps some.
One thing is really going for GIMP: ever since they provided 1-window interface, users no longer find themselves in constant jeopardy of self-inflicted head trauma.
|All times are GMT -5. The time now is 03:16 PM.|