# Last edited on 2025-12-21 22:52:53 by stolfi
# 075 Visual analysis of retracing

  Several details of text and figures on certain pages, such as Zodiac
  from Aries on, were apparently re-traced or added a long time after
  the main text was written and figures were draw. These re-tracings are
  usually conspicuous because of the ink color (darker and grayer
  than the standard ink) and by the places of application (specific figure
  details, and text that apparently had became too faint). 
  
  The agent who did these re-tracings and additions will be called "the
  Restorer". He/she could be the same person who wrote most of the main
  text and draw most of the figures ("the Scribe"), but there are hints
  that the two were different persons.
  
  This note tries to collect examples of apparent restoring and analyze
  them visually.  See Note 080 for a quantitative analysis based on 
  the mulispectral scans of selected pages of the VMS (the Lazarus project).
  
SETUP

  ln -s ../../work
  ln -s work/error_funcs.py
  ln -s work/process_funcs.py
  ln -s work/html_report_funcs.py
  ln -s ${HOME}/lib/argparser.py
  ln -s work/vms_color_image_funcs.py
  ln -s work/vms_linear_gray_image_funcs.py
  ln -s work/bands_table_funcs.py
  
  ln -s work/report_new_annotated_fig.sh 
  ln -s work/report_new_subpage.sh
  ln -s work/report_grab_raw_clip.sh
  
  ln -s work/make_unlighted_MS_page_images.py
  ln -s work/make_composite_RGB_from_MS_images.py
  ln -s work/report_make_raw_image_from_MS_images.sh
  
  ln -s work/MultiSpectral/davis/make_simple_RGB_composite.sh
  
EXTRACTING SAMPLES BY TYPE 

  Starting 2025-05-30, I fetched some page scans ("full size jpeg") from
  the Beinecke page veiwer site,
  "https://collections.library.yale.edu/catalog/2002046"
  and saved them to "~/projects/voynich/FromBeinecke".
  
  I manually extracted from the scan of page f73v (Sagittarius) some
  samples of writing and drawing with normal ink (presumably by the
  Scribe), writing and drawing with the darker ink (presumably by the
  Restorer), and blank parch.
  
  Assuming that the page height is 235 mm, the resolution is 0.066 mm/pixel.
  
  The samples were all manually extracted with Gimp and pasted all
  together onto a single image "samples-by-type/073v/all.{xcf,png}".
  
  Separating the sub-images:
  
    ( cd samples-by-type/073v && convert all.png -crop 'x160+0+0'   retrac.png      )
    ( cd samples-by-type/073v && convert all.png -crop 'x170+0+176' normal-full.png )
    ( cd samples-by-type/073v && convert all.png -crop 'x124+0+348' normal-weak.png )
    ( cd samples-by-type/073v && convert all.png -crop 'x76+0+470'  normal-fade.png )
    ( cd samples-by-type/073v && convert all.png -crop 'x196+0+568' parch.png      )

PREPARING MULTISPECTRAL PAGE IMAGES FOR MULTISPECTRAL CLIPS

  For the pages that have multipsectral imaging, I created versions of the "page.png" images
  from each useful band (frontal two-lamp illumination, {illum=0}), with lighting effects 
  removed and normalized with respect to the "writing" subset:
  
    ??? Move to a separate Note !!!
    make_all_unlighted_MS_page_images.sh

EXTRACTING LARGE SAMPLES OF SAME SIZE

  Fron a few Beinecke page scans, I clipped some samples of text with
  the same square 50 by 50 mm size. For each page, the pixel zise of the
  sample was calculated assuming that the page height is 23.5 mm. Since
  the pages are uneven, the actual dimensions of the samples may be a bit off
  and variable.
  
  These clipped samples were saved to "samples-50x50mm" as files
  "001v.png", "075r.png", etc.
  
REPORT

  A report is being written. See "report/report_src.py" and "report/report.make"