Transforming reference components

If, for instance, my right parenthesis is a reference to the left parenthesis, flipped, are there any known problems with this? 


  • You should rotate it by 180° to avoid a wrong contrast and decompose it on export. Otherwise you should be fine.
  • Nick ShinnNick Shinn Posts: 1,521
    edited February 2015
    Why wrong contrast? Isn’t that only an issue when glyphs overlap?
    Why decompose? Isn’t the idea to reduce file size?
  • Contrast matters if the design isn't horizontally symmetrical.
  • When doing metrics transformations things can get screwy unless the glyph is only flipped vertically and/or has symmetrical sidebearings.
  • PabloImpallariPabloImpallari Posts: 527
    edited February 2015
    I prefer to avoid flipped component whenever possible.
    Sometimes they end up having inverse contour direction.

    Use them at the beginning of the design process. Once you are happy, decompose and correct path direction.
  • attarattar Posts: 209
    But I think mirrored references is perfectly doable with TrueType. I've seen that in FF webfonts (I think I was looking at Franziska).
  • Nick's first question is whether this is a problem, and in storing sources, if it is handy to have a glyph composited, remaining linked to the source glyph, including transformation by whatever useful means the design tool will allow, then do it. For font generation, I'd decompose and correct path directions on anything other than x and y positioned composite glyphs. 

  • When I tried mirrored or y shifted components, the results where not nice...
  • Do you autohint after you decompose everything?
  • Do you autohint after you decompose everything?
    How you do your hinting doesn’t (shouldn’t ?) matter (if the hinting would be decomposed with the outlines ;)
  • Do you autohint after you decompose everything?
    When creating auto hinted CFF fonts in FontLab – particularly when using Adobe’s Auto-Hint script, I have found you get better results if everything is decomposed first. I think this may be due to subroutine building and hint replacement.

  • Deleted AccountDeleted Account Posts: 739
    edited March 2015
    If one is autohinting, it doesn't matter much when the decompositing happens, I think.
    Also, I'm pretty sure cff has nested subroutines so, I don't think you save by decompositing everything. But I'd like to know if cff subroutines use mirroring and flipping matching to compress, or whether the orientation of subroutined curve segment must match...
Sign In or Register to comment.