I have a vested interest in digital imaging, so I will give you as brief a rundown as a geeked-out computer scientist can about the field.
The resolution of an image determines how detailed the scan is. Resolution is commonly referred to in dots per inch (or dpi), where every dot becomes a pixel in the digitized image. Back when your computer monitor weighed 500 pounds a common screen resolution was 72 dpi, meaning that every inch-long line on your display had 72 pixels in it. This has evolved greatly over the years, to the point where an iPhone "retina display" has so many pixels packed per inch (326) that your eye cannot make them out. For printing an image to a book, common resolutions are 150 to 300dpi.
As you can imagine, the higher the resolution of a scan the more pixels will be produced, and the larger the resulting digital image will be.
When scanning, I prefer to scan at the highest possible resolution my scanner will allow (600dpi). The advantage is that I can make enlargements to a photo and know that the result will still look great, even when printed. Digitization is the most important step in the process of preserving your photos and documents as it produces the foundational set of pixels with which you will operate. The better the foundation, the better the house.
Lossy v. Lossless Compression
The main difference between file formats like JPEG and TIFF is the means of compression. Images contain a lot of data, so image file formats counter this by compressing the data within the file format.
As you might expect, formats can achieve greater compression (thus making the file smaller) by throwing away bits of the image. This is how JPEG works. The higher the compression, the smaller the file but the more of the original image is thrown away. This is known as lossy compression, as parts of the original data are lost. The Wikipedia JPEG page has great examples of the same picture saved at varying degrees of compression so you can see what lossy compression does to your image.
TIFF, on the other hand, uses lossless compression algorithms. This means that the original image is completely preserved in the file. The upshot is that the image is never destroyed or altered in an unrecoverable way (barring damage to the file itself). The downside is that lossless compression cannot achieve the same file sizes as a lossy algorithm, so your TIFF files are likely to larger - sometimes much larger.
In the end, it is a tradeoff between file size and image quality, so the real question that needs answering is what is your intent for the images?
- If you are looking to preserve the image in a digital format, the goal should be a high quality reproduction that is as close to the original image as possible. In this case your answer would be TIFF.
- If you are looking to transmit the image over the web (e.g., putting your images in an online gallery) then the goal is to make the file size reasonably small. In this case your answer would be JPEG.
My answer to the question I posed above is actually both, but more on that later when I talk about workflows.
The other issue that needs to be tackled is the question of metadata. These are the bits of information not in the image, but about the image that need to be tracked alongside it. Things like the date the photo was taken, who's in it, who took it, etc., are all metadata. This problem has been around since the introduction of the photograph, and the time-honored solution was simple: write it on the back of the photo!
In the digital realm, the "back of the photo" is actually a space within the file format itself. This is a relief because wherever a photo may travel, so too travels its metadata. There are several different ways one can store metadata in an image (IPTC/IIM, Exif, XMP, etc.) with varying degrees of support. As long as you have a decent image editor you shouldn't have to worry about about any of them, as it should handle them all.
Usually the program used to save the image should give you a place to edit or add metadata to the image before saving. A quick Googling for metadata editors also gives a list that let you change the data after the fact. These editors should have fields marked clearly for the input of various metadata. The editor, then, will "do the right thing" and tag the metadata you enter appropriately.
So now that you've had a drink from the image format firehose, the question is how best to incorporate the knowledge into a reliable workflow? In other words, what's a good process for going from physical image to digital representation?
This is my workflow and it works well for me, though that's not to say it is the best workflow per se:
- Scan the image full-color at 600dpi, turning off all post-process steps in the scanning software (auto-align, color correction, etc.) If you have a digital camera you can also use it to "scan" the image by taking a well lit photo (as a RAW image). Given the condition of the source document, this can be a tricky step.
- Save a copy of this file as a TIFF using (lossless) RLE or ZIP compression. This is my "RAW" copy. (If I took a photo, that RAW photo is my "RAW" copy.)
- Open the file in Adobe Photoshop (disclaimer: I am Photoshop programmer) or your image editor of choice.
- Straighten the image, crop it, and add all relevant metadata. I would also touch up the image at this point if necessary or desired.
- Save a copy of this image as a Photoshop document. This is my "PSD" copy.
- Reduce the image size and save a copy using Photoshop's "Save for Web" feature. This will produce a compressed JPEG version of the image suitable for the web or email. This is my "JPEG" copy.
- Archive all three versions of the image: raw, psd, and jpeg. This is done by burning a DVD, saving to an external hard drive, etc.
When all is said and done, I have three versions of the image at varying stages in the workflow. The most critical step is the first one, as all other steps can be reproduced from that raw image if need be. Adding the metadata to the PSD copy and using Save for Web ensures that the metadata will be copied into the JPEG version of the file as well.