The Zombie Stack Exchanges That Just Won't Die
I have many photos taken on analog that I'd like to scan to digital form to preserve them. Some will be scanned in using a flatbed and some using a slide scanner.
What format (JPEG?) and resolution should I be aiming for ? I see resolution on digital cameras has evolved quite fast. We've gone from 1 megabit to 15 megabit and that might increase further. Should I use just JPEG or use a non-lossy format in parallel to accomodate further developments ?
Furthermore, if these are family photos or the like, should anything be done to adjust colours or would a simple scan be fine ?
James Poulson
When scanning analog photos, the JPG format should be sufficient (with a "high" quality setting; i.e. low compression). There's little point in storing it in a lossless format. The reason why DSLR cameras save photos in a Raw format is to preserve as much data as possible from the original sensor for postprocessing. But since the analog photo is already developed, there's no point in trying to squeeze out detail that doesn't exist.
As for resolution, I've actually found that 300 dpi is quite enough. Any more would be going beyond the "resolution" of the physical photo paper, and wouldn't really add any more detail. (And 3000 dpi for slides)
You shouldn't need to adjust the colors, either (except to make sure the scanner is scanning the photo correctly in the first place).
You may want to digitize to a non-lossy format if you're going to do any processing (e.g., cropping or color correction to adjust for scanner issues) on the image. Editing a JPEG image can result in further degradation. A big archive would keep a permanent lossless image for the same reason, but that's most likely overkill for personal archiving. Once you're satisfied with any adjustments you've made, you can archive them to a high-quality JPEG file.
I have found that a DSLR with a low distortion macro lens will do a much better job of digitizing photos than any scanner that an ordinary person can afford. Scanners are fine for the big picture (and better than nothing) but for preserving fine details, the DSLR does a much better job. The secret is lots of light. The best way is to use a copy table/stand, making sure that the arm holding the camera is rated for its weight.
As to the color, I generally try a few different adjustments to the color to see what happens and use whatever looks best - which is sometimes not making any change at all.
For file format, JPG is fine, but use a high quality mode.
When you want to preserve something for a long-term period, you should always consider the following topics:
1 - you don't know what you'll do with it in the future: if you have a precise idea in mind of all the future uses, you can choose how to store the images according to that uses. But very frequently you (or somebody else) will want to make different uses of it, and then will blame who "preserved" it for not taking enough care.
2 - so, if it's within your possibilities, put as much effort as you can in doing it. If you have enough media storage, use as much resolution as possible. 300 dpi might be good today, but also might be to few in the future. One day someone though that VHS was more than enough to store video, and today bluray is good, but it's not that much, is it?
3 - if some file format is widespread enough today, probably it'll be easier to find something about it in the future. Everybody uses JPG, so go the same way. With as much quality as possible (perhaps a 0.98 quality).
4 - If you were dealing with high-quality photos, you would have to worry about color calibration. Every equipment has it's way to represent color, and so your photos have one color, that will be transformed slightly by the scanner, and your computer monitor will also show some different colors when you view them. So, yes, if you're not using that high-end calibrated scanner, you could do some color adjustment, or save some color reference like the color profile...
5 - ... but I think you're talking about family photos, not spending that much to scan them, etc. So you could scan some color reference to help someone in the future, why not? Buy some scanner calibration sheet (for example, this one, scan it, and store a copy of it with each folder you're organizing your photos. That way, in the future, someone who will be using your pictures will have an idea on how your scanner represented the magenta, cyan, yellow...
The ideal format for preservation at this time is TIFF with 24bit color and no compression. PNG with lossless compression is a reasonable alternative. JPEG using the highest quality setting (minimal lossy compression) will serve many personal preservation needs and some professional, well enough. Adobe DNG for preservation coupled with JPEG is worth considering, and is a method used for high-quality digital photography, but could also be used for scanned photos.
The "extra" questions about resolution and color adjustments are probably best answered separately. Nonetheless, they are important, and so briefly:
400 pixels per inch (ppi) scanning resolution is sufficient for most printed photographs. 600 ppi might be necessary to capture the grain in certain situation. Approximately 3000 ppi is typically sufficient to capture the full detail of a photograph from 35mm film. Only use the native optical resolutions of the scanning device. Color management using ICC color profiles and calibration techniques will help to achieve the best possible results.
The hardware used to do the scanning plays a huge role in the quality of the results.
Digging into the Details:
The ideal format for preserving analog digital photos is one that:
TIFF largely complies with the above. More specifically, TIFF with 24 bits of RGB color or 8 bits grayscale for black and white images, without compression. It has proprietary origins but is open and probably more widely adopted than any other format making it a de-facto standard. Like most formats, there are many varieties of TIFF. TIFF without compression is most appropriate. TIFF is rarely used with LZW lossless compression within the preservation community, because LZW has multiple commercially held patents associated with it, making it undesirable.
PNG is a viable preservation format that supports lossless compression that is free of patents (unlike LZW). If TIFF had not come first, PNG might be used more widely for preservation than it is.
JPEG does not comply because is uses lossy compression. Every time a lossy compression is resaved, the visual artifacts of the lossy compression are compounded and the quality of the image degrades. Some software applications, such as iPhoto, mitigate the effects by keeping the original so the option always exists to revert to it. Technically, the JPEG format supports lossless compression as well, but this feature is so rarely implemented in software that it cannot be depended on in the preservation context.
JPEG2000 complies if used with only lossless compression (a.k.a. reversible), though it is relatively a very complicated format. It is used extensively by some large preservation repositories, but is not a good choice for the average user.
RAW formats tend to be proprietary and therefore not as reliable in the long term. RAW formats store the greatest amount of visual data because it represents what the hardware is actually capable of, but are not supported widely by viewing applications. Adobe Digital Negative (DNG) is, however, a relatively good option in this realm. Pair Adobe DNG with JPEG and you have a rich master and an easy to use copy, in that order.
BMP is a proprietary format controlled by Microsoft that is deeply integrated into the Windows operating system. It was not designed for interoperability. Therefore, it has not been used widely for preservation.
Finally, this topic is one that has been examined thoroughly on an ongoing basis for a couple of decades in the digital preservation realm and there is a ton of information available on the web if you care to explore it further. I've tried to give a concise answer (ha!) without compromising preservation ideals. The external links of provided barely scratch the surface.