DAACS Image Processing Instructions

Written by Chris Jessee, March, 2004.
Charlottesville: DAACS, 2004.

Source Material

DAACS digitizes the following types of source material:

  • Line Drawings
  • 35mm Slides
  • CAD files
  • Physical Artifacts

Process Overview

Each source has its own unique characteristics that require special consideration during digitization, while the basic process remains the same. The process is as follows:

  1. Digitize the source material.
  2. Record metadata about the image and material.
  3. Archive images to CD-R discs as LZW compressed TIFF files.
  4. Store duplicate compact disks in an offsite location.
  5. Color correct, crop, adjust and otherwise batch process images for delivery.

Digitization Rules-of-Thumb

Regardless of the capture device, a few rules-of-thumb apply to most source materials:

  • Record the image at a resolution that can capture the smallest significant detail of the source material.
  • Turn off all automated filtering and exposure correction in the capture device, in order to capture the greatest possible dynamic range. Filtering processes are more accurately and consistently performed with image-editing software, rather than scanner software.
  • Capture images at a 24-bit color depth. If a lower color depth is needed for delivery, process the image down with image-editing software. The color reduction software that is built into capture devices is typically lower quality than desktop image-editing software.
  • Capture a color target in the middle of each set of images to assist in color correction and normalization between sets.
  • DPI is a largely irrelevant scaling factor. The total number of pixels captured across its height and width is a far more important measure of an image's usefulness. A 300 or 600 DPI may not be adequate if the overall pixel count is too low.

Processing Instructions

Line Drawings

Line drawings are typically scanned on a flatbed or sheet-fed scanner. If the drawing is double-sided with black lines on a white or light field, it is helpful to use a dark sheet of paper behind the drawing. This reduces the back image showing through on the front when scanned.

35mm Slides

35mm slides are typically scanned on specialized slide-scanning hardware or a flatbed scanner with an overhead transparency adaptor. All available detail can be effectively captured from a slide by scanning to a 4000 x 3000 pixel image. This will yield an approximately 35MB file.

CAD files

CAD files originating in Microstation, AutoCad, or similar software require no digitization, as they are already digital files. A variety of pixel and vector files can be saved out of typical CAD software. It is good practice to archive the original CAD file, an exported DXF file, and any derivative image files such as TIFF or GIF to a CD-R. Should the software become unavailable, the derivative files will still be accessible. Given the nature of the content of CAD files, they are best delivered on the web as GIF images.

Physical Artifacts

Physical artifacts can be digitized with a flatbed scanner, a digital camera, or a film camera in combination with a slide scanner. Slides are a poor choice for archival images because of their limited dynamic range and the additional costs/steps involved with film, processing, and scanning. A digital camera often yields better results than a scanner but requires additional work to ensure proper lighting. Lighting setup is beyond the scope of this document; please see the Additional Information section below for more information.

It is often practical to photograph and take meter readings from an 18% gray card. This provides a neutral point of reference to adjust the camera's white balance and compensate for color shifts introduced by lighting. Digital cameras used for generating archival images should be capable of saving an uncompressed TIFF file. JPEG files are not suitable for archival images. If a suitable camera cannot be obtained, the iterative degradation problems of successive JPEG compression can be reduced by first saving the camera's JPEG image as a TIFF file. This may be acceptable quality in some situations.

Image Processing

Image processing can be divided into two distinct types of tasks, manual and automated. Manual processing requires human intervention to crop, straighten, color correct, and retouch images. The production rate of manual processing can be dramatically improved though the use of keyboard shortcuts and practice. It is important to limit the number of people performing manual processing on a set of images, since more people increase the degree of variability introduced into the set.

Automated processing involves writing scripts or macros to iterate through a set of images. A variety of filters, adjustments, and transformations are applied to the images. the resulting files are saved with new names and in a different directory location. It is good practice to select a small, representative sample of images for use in the development of batch processing scripts. It often takes a dozen or more test runs to achieving the desired settings for a script. Reducing the sample set to a small number of images substantially reduces the testing cycle and thus reduces script development time. See the Software section, below, for additional information on software choices for batch image processing.

Metadata and Archiving

Metadata

Two types of metadata are required for still images: technical and preservation. Preservation metadata contains administrative and structural information. DAACS has extensive metadata methods already in place so this topic won't be covered in this document. See the Additional Information section for links to image metadata standards.

Archiving

A good rule-of-thumb for archiving digital image files is to never have fewer than two copies of an authoritative source image at any time and never have the copies in the same place at the same time. Use LZW-compressed TIFF files as the authoritative source files. LZW is a lossless compression and can save significant amounts of disk space.

The most economical means of archiving is to burn the images onto multiple CD-R discs. Kodak Gold and Mitsui brand disks are the most reliable. Rotate the media at least every 10 years and more frequently if you cannot afford to loose the data. Avoid labeling the discs with pens or adhesive labels (the glue and ink will eat through the disc and damage the data).

Burn CDs in ISO9660 format and use the Joliet naming conventions. The Joliet extension to the ISO9660 standard allows up to 64 character file names (including spaces) making meaningful file names convenient. Burning CDs in proprietary formats, such as UDF, or on CD-RW or multi-session discs will render them useless in the long term.

Technical Infrastructure

Hardware

Scanning and image processing tasks put extreme demands on computer hardware. The majority of companies focused on imaging choose Apple Macintosh computers because they are configured and optimized for these tasks. Here are a few of the features that make Macs well suited for imaging:

  • High quality monitors with integrated Colorsync calibration software
  • Built-in, high-speed networking (10/100/1000BASE-T Ethernet and 54mb wifi)
  • Fast, high capacity memory (up to 8GB DDR 400MHz PC3200 SDRAM)
  • Fast graphics (8X AGP with a variety of graphics cards available)
  • Fast data path between system components (1GHz front side bus per processor)
  • Dual 64 bit processors
  • Large hard drive (250GB Serial ATA)
  • Fast standard peripheral connections (FireWire 800, 400 and USB 2.0 ports)
  • Low cost, high capacity external storage array

There are over 66,000 PC viruses and growing daily, yet only 80 Mac viruses. The majority of these only affect older Mac operating systems (version 9 and earlier). PC support costs are four times higher than for Macs, largely due to virus related problems. A few days of downtime or lost work due to a virus infestation turns a "bargain" PC into a very expensive mistake.

Non-Mac desktop computers can be configured to do the job, but at higher initial and long term costs. The first and most expensive addition a PC will need is a high quality monitor. LaCie offers a number of options including Photon 20Vision and Electronblue IV. LaCie also offers an excellent color calibration solution for the Mac and PC, the BlueEye Vision.

Numerous high quality flatbed scanners are available from Hewlett Packard, Epson, Microtek and others. A professional-grade scanner like the Epson Expression 1680 is a good choice as it handles both flat art and slides in one device. Several of the resources in the Additional Information section contain scanner selection criteria.

Digital cameras suitable for archival imaging change almost weekly. See the Digital Camera Reviews link in the Additional Information section for assistance with camera selection. The main selection criteria for archival imaging cameras is as follows:

  • Save an uncompressed TIFF file.
  • Contains a high-capacity storage device, such as a micro drive.
  • Record images at a resolution adequate to capture the smallest significant detail of the source material.
  • Fast, easy means of moving images from camera to computer, such as USB 2.0 or FireWire.

Software

The standard software for image editing is Abode Photoshop. It has all the tools needed for manual and automated image processing. The only significant drawback to Photoshop is its slow performance in batch processing. Graphic Converter (free, bundled with new Macs) is a much faster solution for batch processing as it doesn't display the image during processing but simply loads it into memory. MediaPro by IView is another useful tool for sorting image catalogs and recording metadata. MediaPro quickly builds web galleries that are useful for quality control. It also exports its field data as text for inclusion in XML documents or a database. Photoshop, Graphic Converter, and MediaPro are all scriptable via AppleScript. The Smithsonian Office of Imaging and Photographic Services uses Applescript as part of its image management system, scripting a variety of desktop applications as well as a FileMaker database. See the Apple Computer and Smithsonian link in the Additional Information section below for details.

Physical Environment

Of equal importance to the software environment is the physical environment used for editing. Consistent lighting is the first priority. This usually involves a room with no windows, to ensure that the lighting is the same throughout the day and from one day to the next. When viewing flat art or artifacts and making on-screen color adjustments, the room lighting dramatically impacts the perceived color objects. Light fixtures should be lamped with bulbs having a 5000K-color temperature. These bulbs are readily available at hardware stores under the names "daylight balanced" or "full spectrum" lights. Another simple change that can markedly improve color editing is setting the monitor background to medium gray with no color saturation. This provides a neutral base for image editing.

Reference Resources

  1. "Moving Theory into Practice: Digital Imaging Tutorial." Cornell University Library's on-line tutorial. http://www.library.cornell.edu/preservation/tutorial/index.html
  2. Moving Theory into Practice: Digital Imaging for Libraries and Archives. ISBN: 0-9700225-0-6. Available for purchase at http://www.rlg.org/pub.html
  3. "'Automatic Exposure' -- Technical Metadata for Digital Still Images," an RLG-led initiative. Home page is http://www.rlg.org/longterm/autotechmetadata.html
  4. UVa Metadata (information and reports). http://www.lib.virginia.edu/digital/reports/metadata.html
  5. Library Preservation at Harvard: Digitization Resources. http://preserve.harvard.edu/resources/digital.html
  6. Luminous Landscape. A photographic resource site. http://www.luminous-landscape.com/
  7. "Image Archiving at the Smithsonian." http://www.apple.com/pro/photo/jones/
  8. EXIF and Related Resources. Resources and information about the exchangeable image file format. http://www.exif.org/
  9. Kodak Q60 Color Targets product information page. http://www.kodak.com/global/en/professional/products/films/catalog/kodakProfessionalQ60Targets.jhtml
  10. Apple Computer, Inc. homepage. http://www.apple.com/
  11. AppleScript product information page. http://www.apple.com/applescript/
  12. LaCie home page. http://www.lacie.com/
  13. Digital Camera Reviews and News. http://www.dpreview.com/
  14. GraphicConverter product information page. http://www.lemkesoft.de/en/graphcon.htm
  15. iView MediaPro product information page. http://www.iview-multimedia.com/products/mediapro/index.html
  16. Compact Disk FAQ. http://www.cdrfaq.org/
  17. "Burn It." Another CD FAQ, dated but still useful. http://jefferson.village.virginia.edu/~cj8n/doc/burnit/