Address:
Unit 1, 10 Surrey street
PO Box 51248, Tawa
Wellington 5249
P:+64 4 232 9396
F:+64 4 232 9399
E:info@micrographics.co.nz
LOGIN

2. Digitisation

How we positioned the Book

The book could not be opened to an angle beyond 120°. We chose to use our angled book cradle, which can adjust its supports from 45° through 170°. We used closed-cell plastic foam (LD45 black Plastazote supplied by Dunlop Foams ) covered by washed unbleached calico (supplied by Triptych Conservation Services).

We positioned the book on the book cradle in reverse so it appeared back from the centre of the lens. The primary reason for this was to ensure ease of handling the rare book. We did not have to reach ever the top or sides of the upright side of the book cradle support to adjust the pages. The upright rest was secured firmly to ensure it could not move or change angle. This also meant that the camera could look more into the centre of the book, allowing better capture of the gutter margin and maximized lighting potential in the gutter area. We anchored snake weights to the upright book rest to control and prevent errant movement of the upright page during digitisation. We went to great lengths to ensure the snake weights did not come into contact with the gold leaf in fear of snagging or catching and the same went for the page edges.

Handling Difficulties

The vellum had various degrees of buckling, causing wide variances in page height. Due to bowing of the pages the book was effectively thicker in the middle than on the edges. We considered and employed a number of ways to overcome this - where the corners were turning up we chose to use flimsy mylar strips held down very gently by tri weights outside the scan area. When the sides were "falling down" we "propped" up the edges of the pages by careful insertion of small pieces of archival foam placed in between the end papers.

In all cases great care was taken to ensure the horizontally-lying pages were supported across their entire dimension. The back cover was also supported in a similar fashion and there was a measured gap between the two book cradle "rests" to minimise pressure on the spine.

The method of employing mylar strips can be seen in the picture below. To ensure there was confirmation in the digital image we added a small target piece of paper just off the page citing "MYLAR" to explain the subtle difference in colour.

We chose to position Plastazote foam in the end papers to minimize risk to the illuminated pages. The end papers were used as not only were they not contemporary with the text block, they also were of a more robust nature. We used a combination of mylar and Teflon bones to hold down the edge of the page where it was pulling up, the Teflon gently resting on the page on top of the mylar during the digitisation. It is possible to "retouch" out these in the final images, perhaps in the event of a republication; however we have not done this as we have tried to show the images as original as possible.

It was extremely important to measure the height of the page from the base at every scan. As we were "propping" up the edges of the book to maintain a relatively flat and supported page this had the effect of changing the distance between the lens and the book at every page turn. We measured this height at every page turn and compensated for this at the scanning stage by changing the parameters in the scan program which regulate the distance between lens and target.

As we used an angled book cradle it was decided to scan all odd pages (rectos) first, working from the front to the back of the book, then to turn the book around when the last page was reached and scan all the evens (verso's) from the back to the front of the book. In this way book handling was minimised.

Scanning of the 222-page book took approximately 30 hours.

Whenever the book was removed from the camera i.e. end of day, when it was put away, we carefully marked the area of the book cradle where the book lay with pieces of Plastazote foam to indicate its position for the next day. This ensured that setup time would be minimised, ensuring less handling and less exposure to light.

Great care was taken at the turning of each page with paper triangles to ensure:

1. that pages were not damaged, and were always well-supported during movement

2. that gold leaf or text was not "snagged" or smudged by gloves or triangles

3. that upright pages were held gently in place by snake weights and horizontal pages were not left unsupported underneath or above, as we progressed through the volume.

Workflow for Image Capture

The workflow was undertaken by a team of two digitisers.

The first digitiser was responsible for the setup and arranging of the book on the book cradle, and the turning of each page and consequent alignment of each page using combinations of snake weights, foam, mylar and bones, as discussed in Handling Difficulties.

The second digitiser is responsible for the operation of the scanning software and the Quality Control of the digitised images. Due to the power of our computer systems and speed of our network it was possible to undertake Quality Control and post-processing on one computer while performing scanning on another.

The second digitiser was also used for setup and arrangement of the book on the book cradle in circumstances where four hands are better than one. However, generally, separating these two distinct tasks made the workflow easier and faster. By separating the responsibilities it made it easier for the digitisers to concentrate on their individual tasks and provided more accurate results with less risk. We decided to capture the image of the page plus a further 30mm around the edges to allow for background to be seen in the final master images, once post-processing was complete.

The workflow of the image capture procedure contains mainly five steps for the First Digitiser and Seven steps for the Second Digitiser, each of which are presented here chronologically:

First Digitiser's role

1. Set up book/page turn/check page supports on upright and horizontal book rests.

2. Decide if Mylar needed to stabilize loose page. Measure height at seven points to (as a guide try to keep all points within 5mm height range from base).

  • Left back
  • Left middle
  • Left front
  • Right back
  • Right middle
  • Right front
  • Centre front

 

3. We have found the best method of achieving accurate average focal point area was by checking the front middle of the page. We paid particular attention to keeping the central text block flat because the wide margins of the page generally contained blank space. We also could not apply any pressure to the inner curvature of the pages for fear of damage.

4. Enter height information into the scanning software.

5. Check the name of the scan to be performed is correct (usually set as a generic filename and automatically incremented by two (rectos), or decreasing by two (versos) and that it is saved in the "to be QCed" (QC = Quality Control) folder.

6. Click the scan button.

Second Digitiser's role

1. Open up the last scan digitized in the "to be QCed" folder in Adobe Photoshop.

  • Check the image thoroughly to ensure there is no corruption. Things to look for are telltale thin white lines in the image which could indicate a problem reading the tiff data, or blocked pixel elements in the CCD.
  • Check to ensure there are no dark vertical lines in the scan. This could indicate someone walking too close to the scanner during operation, causing a drop-off or change in lighting.
  • Check to ensure the clarity of the image - check the sharpness of the image. This could indicate the wrong height was entered into the scanning software, causing the lens to be too far away or too close to the object being scanned. The most important information tends to be the actual type on the page and where we have had problems with uneven pages we have tended to ensure that the type is clearly readable and the edges appear softer.
  • Check to see if the highlights and shadows are well represented - there should be no plugging up of the shadows or blowing out of the highlights. Open the levels histogram (Image:Adjustments:Levels). There should be no clipping in the image and the triangles indicating the white and black points should show a falloff of data before they reach the triangles. If the data goes right up to the triangles chances are that you have too much contrast in your image and have lost detail.
  • Finally check that the image is at the correct resolution. It is easy to incorrectly set the pixel-per-inch ratio in the scanning software and if this happens, while you have the correct amount of pixel information, the final size is wrong and you could end up with a file that appears to be either half or twice the original size, or any ratio in between. The image needs to be at actual size. If there are problems in any of these previous checks you will need to make a decision whether to rescan the page again.

 

2. Using the [measure] tool draw a line along the outside page edge to determine the angle.

3. Select Image:Rotate Canvas:Arbitrary… The angle the book is out of square from 90° will be displayed here (i.e. 0.12° CCW) Rotate this angle to deskew if necessary. The odd pages will have to be rotated a further 180°.

4. Move the zero point to the outside top corner of the book then draw guides to 20mm outside this edge. The book was 190mm x 240mm. Using the [marquee] tool we had preselected a 230mm x 280mm area. When we positioned the marquee on the guides we had a selection area which gave us 20mm around the entire book and included at least 20mm of the opposing side page.

5. We then selected Image:Crop to allow the image to be cropped into this new declared selection area.

6. We then applied Unsharp Masking of 95%, 2 pixel radius, 2 level threshold to the image.

7. The final image was then saved into the [QCed] folder on the original scan machine. With practice we found that the second digitiser was able to achieve all his work in the time it took for each image to be scanned, so that the Quality Control and Post-Processing was achieved in almost the same time as the capture. As these were relatively small images for the Cruse camera (6500 x 5500 pixels 24bit RGB - 103MB) we were able to post-process and write the images in this time. This may not be achievable on larger images, especially as the Cruse can scan up to 15,000 x 10,000 pixels 48bit RGB -900MB!