Digitizing on a budget

Professional digitization equipment can be extremely expensive. Even mid-range equipment can be out of reach in many situations, due to price, shipping, or transport restrictions.

In tight situations, where the immediate digitization of artifacts is more important than achieving a state-of-the-art image quality, there are more accessible options. Below are some ideas and resources to start a DIY digitization studio.

Note: this document is exxclusively focused on still image capture. I don't have enough working experience with audio, video, and other born-analog media to provide advice on those.

Choosing equipment

The choice of a digital camera and/or scanner entirely depends on your budget, and a long discussion about the various options is out of scope for this document. We are assuming that you have a restricted budget, some technical expertise or someone able to help, and some creativity.

Camera

A compact point-and-shoot camera is probably the best compromise between price and quality for who cannot afford a DSLR camera. They are also very portable. There are plenty of them in the used camera market.

Canon compact cameras are of very good quality, and present an additional advantage: a community of enthusiasts developed a custom software tool called Canon Hack Development Kit (CHDK), which can be loaded on the camera's SD card and enables several features that allow a fine control of the camera, such as exposure and white balance, necessary to get a professional quality out of your equipment. It also allows to capture images remotely by controlling the camera from a laptop or workstation, and directly downloading the images via USB cable.

If you are able to set up CHDK, browse around your neighborhood's used camera shops or online for bargains, check the specs and CHDK compatibility in this model cpmpatibility page, and get the best one that you can afford. You won't likely need any more than 16 megapixels. Rather than just the resolution, look for larger and better quality lenses (e.g. Carl Zeiss).

If CHDK is not for you, you can always choose a camera with enough manual controls built in; in which case, other equally good brands are available.

If all else fails, a cell phone camera can still work.

Scanner

Flatbed scanners are the best choice for flat objects, as they yield the best quality. They have, however, a limited range of mateirals that they can capture; if you have a mixed collection, you will end up having to buy a digital camera anyway.

Digitizing your materials

Small flat items

Documents, postcards, posters, letters, etc. are best digitized with a flatbed scanner. This yields the best quality and consistency.

If a flatbed scanner is not available, a digitization setup can be obtained with only a few items:

• A clean, scratch-free glass plate;
• A digital camera (even a relatively recent cell phone)
• A tripod that fits the camera at hand
• One, or better, two sources of light

The artifact is placed under the glass to keep it flat. The glass must be larger than all the items you plan to digitize.

The environment around you should be as dark as possible to avoid objects being reflected into the glass.

The light source(s) must be placed at an angle so that the glass doesn't reflect the glare into the camera, and the camera doesn't cast a shadow on the subject. They should also be placed far from the subject enough to ensure an even lighting. The farther the light source, the more even the lighting will be, but also the weaker. Some adjustments and trial may be needed.

If the items are already quite flat, it's best not to use the glass, for better image clarity.

The camera should be as parallel to the surface as possible. If the item and glass are placed flat on a table, the camera should be straight overhead. Some margin around the object should be left for later cropping, but the camera should not be too far, as not to lose too much resolution. If possible, organize your materials by sizes, so if you need to move the camera, you do it as little as possible.

If you have a camera with manual or semi-automatic exposure controls, you should use a relatively high f-stop (aperture) and relatively slow capture time. It is best to lock the focus once set, if you have the option to. If you have a delay timer, you can use it to minimize the chances of shaking the camera when you press the shutter button; or use a remote trigger as described earlier for the CHDK setup. If you have a white balance function, use it to adjust the color so that white appears white. This yields better quality pictures than adjusting the color in post processing.

Remember to scan front (recto) and back (verso) sides, unless you need to save time and the back side is completely blank. Even small inscriptions, stamps, labels, etc. can be valuable metadata to track the provenance and context of the item.

Shoot at the highest resolution that your camera supports. Normally, storage is less of a concern for smaller collections, but you should revise your settings after some captures, predict the required storage space, and adjust your output resolution accordingly if you are tight on storage resources.

If your camera can shoot in RAW format, use it. That allows for post-production color and exposure fine-tuning that don't affect the image quality significantly.

The image resolution for text-only documents should be enough to allow good quality OCR (Optical Character Recognition). These documents can be converted to grayscale (black and white) which are usually 1/3 the file size of color images. More care and a higher quality should be used when digitizing images.

Books

Books are more complex to capture than flat papers, as the book binding bends the pages and the images most likely need to be geometrically corrected after the capture, with inevitable loss of quality. Older bindings are also susceptible to breaking if they are forced open too far.

For ones with a modest budget and little time, some economic, portable book scanners with integrated image enhancement software are available for a few hundred USD. These don't need a book cradle; however, I haven't used them personally and can't guarantee their quality. As with color correction and other adjustments, the best image quality is obtained by making geometric corrections before shooting. This is most important when only modest means are available, as the room for post-production adjustment is severely limited by the caturing equipment.

A better option is to use a book cradle, which helps keeping the page open and as flat as possible without forcing the spine and damaging the book. "Professional" grade book cradles can reach astronomical prices, but good results can be also obtained with a home-built setup. The DIY book scanner forum is an excellent resource to build scanning cradles for all budgets and all skills.

Capture all the pages, even blank ones, to keep the correct numbering sequence. Choose a file naming convention with a zero-padded numbering suffix (e.g., 0001, 0002, etc.), which helps keeping the original order and allows the automatic generation of an ordered tree structure when the file list is copied to a laundry list.

Large flat items

Large paintings or panels can be challenging to digitize because of the difficulty of creating an even lighting on a larger surface. More, and more powerful, lights are needed for these objects. In the absence of good artificial lighting, the sun is an excellent option: choose a cloudless day, and make sure that the light doesn't fall on the object at a too sharp angle (raking light), which could reveal the roughness and imperfections of the surface (especially on painted canvas) to the point that they affect the readability of the image.

Some surface texture may be desirable to show, as it's part of the work; at times, additional images are taken with raking light for detail and conservation purposes. Therefore, getting the light angle right is quite important. Some detail shots may also be desirable. All images of the same object should be named and/or grouped under the same folder in a way that makes it easier to catalog all the images together as related to the same artifact resource.

If you are shooting multiple objects outdoors that take a while to set up, check the exposure and the white balance every now and then as the sunlight changes.

3-D objects

Three-dimensional objects should be captured at multiple angles. The best way to do that is to keep camera and lights static, and rotate the object. Any kind of spinning plate that can support the object does the job.

Secure a round, flat panel with a neutral color on the plate, marking a dot on the center of rotation. Center the object on top of the dot. Place this setup in front of a neutral-colored wall or panel.

The number of images you want to take for each object depend on the complexity of the object and the storage you can afford.

Post-processing and archival

This step is necessary to get a final quality archival image. A detailed explanation of image post-production techniques and workflows is not in this document's scope, but suffice to say that these steps usually involve the following:

• Color balance adjustments
• Exposure and tone adjustments
• Removing shape distortion (deskewing)
• Cropping
• Retouching, e.g. to remove extraneous details, backgrounds, etc.

For each digital capture meant to be preserved, at least two files should be kept:

• An archival master file, which retains as much information as possible from the original capture. This file is adjusted to the desired tonal range and is usually kept at its original bit depth (most likely 16 bits per channel). It should keep adjustment layers, additional channels, etc. so it can be modified further if ever needed. The image is not cropped. This file acts as a reference, as the "source of truth" for other files derived from it, that can be used for a variety of purposes.
• A production master file, obtained from the archival master. This file is cropped to its final presentation frame, has all layers and transparencies flattened, is converted to 8 bits per pixel, and retains its original resolution. This file is used to generate derivatives via automated processes.

If one can afford it, both files should be preserved, as they provide the best guarantee of recovering a lost or damaged derivative (which is not preserved) without having to re-digitize the original. If storage is tight, the choice is to only preserve one of the two, and assign it both preservation and production master roles: by preserving only the former, one retains the most information and the ability to adjust the original image within a reasonable range without loss of quality, but manual work and human review are needed to regenerate a derivative, which won't likely be identical to the lost one; by only preserving the latter, a derivative identical to the lost one can be generated automatically, but further adjustments to the image will result in quality loss.

The original RAW file is usually not preserved, as it adds significant storage volume.