Subject: FindMatrix                                               Author: Bloody

(The intro below is an excerpt from the /FINDMATRIX documentation in CLI.txt)

Introduction: the issue with huge Matrix files

  The purpose of this command is to allow using separate matrix files per font
  (or per DVD/series) in order to avoid growing huge ber matrix files with
  hundreds of fonts which would eventually lead to erroneous output, like
  confusion between J<>j, o<>O|0, V<>v etc. (and even worse).
  Attempting to 'fix' such a giant matrix can be quite difficult since it does
  not necessarily contain 'wrong' glyphs, but simply too many of them.

  To vamp up such a matrix, one could tighten some of the Advanced OCR options
  (see --ocr-char-*diff* options), e.g. by raising --ocr-char-diff-sensibility
  to 1000, but that in turn would require adding otherwise equal chars multiple
  times just because of subtle differences, which is not only more work but also
  increases the size of the matrix even more, requiring even tighter settings,
  and so on.

  Large matrix files also have the disadvantage that they cannot be split into
  multiple small matrix files because there is no way to tell which glyphs
  belong together (same font). Once a mess, always a mess.

  With the /FINDMATRIX command, matrix files can be kept separate (per-source),
  searched for, sorted out (by finding & removing 'doubles'), more easily fixed
  (or simply deleted), and even be shared between friends to build a collection.

  Another advantage of using small matrix files instead of huge ones is that
  subtitle conversion goes much faster, which can significantly improve batch
  conversion speed.

  Note that using separate matrix files also has it's disadvantages, regarding
  administration effort, i.e., having to run a /FINDMATRIX command before
  converting subtitle(s). At the end, one might still choose between using one
  large matrix or several small files (or even a combination of both).

GUI Help: using FindMatrix in GUI mode

  In essence, the automatic Matrix search will scan a number of unique glyphs in
  the selected subtitle (default: 60), then start the search by checking if a
  minimum number of those glyphs (default: 40) can be found in any Matrix file
  to consider the Matrix to be a 'good' one for the selected subtitle.

  If no suitable Matrix was found, a new (separate) Matrix file will be used for
  the font glyphs of the selected subtitle.

  The automatic matrix search can be manually started at any time after opening
  a DVD subtitle for conversion (.ifo/.vob or .idx/.sub), by clicking on the '>'
  button on the very right.

  The "Auto-Apply" checkbox can be used to suppress the small result popup
  dialog and instantly apply (load) the Matrix (either a 'good' one or a new one
  if nothing was found), saving a mouse click.

  The "Auto-Find on Start" checkbox can be used to entirely avoid any further
  mouse clicks; you don't even need to select/clear the Matrix file anymore.
  Just select a subtitle to convert and then click on "Start" to auto-find a
  'good' Matrix and then auto-proceed to the subtitle conversion.

  If you don't wish to use the automatic Matrix search feature, just see that
  the "Auto-find on Start" checkbox is unchecked, and then simply ignore the
  remaining FindMatrix buttons/widgets and proceed as usual.

  Note: the name of the current Matrix is now also displayed on top of the "Add
  new characters" window so you'll always know which Matrix you're working with.

Hint: avoiding the creation of micro Matrix files

  When converting new subtitles (i.e., from a 'new' DVD), try to pick a larger
  subtitle first (for that subtitle language), e.g. by opening the 'main title'
  first, so if a 'good' Matrix could not be found, a more complete new Matrix
  will be created from the larger subtitle instead of a possibly very tiny one.

  This makes it more likely to find the newly created Matrix in the next Matrix
  search for another subtitle that uses the same font. Had you first chosen a
  very short subtitle from a small 'bonus' clip or a tiny 'forced' subtitle, the
  created Matrix could be so tiny that it would be of little use if it doesn't
  meet the requirements for a typical Matrix search (40 out of 60 glyphs).

For further (technical) infos, please refer to CLI.txt (under "/FINDMATRIX").
