Bad Private Dictionary

mauro sacchetto
mauro sacchetto Posts: 353
edited December 2018 in Font Technology
In modifying a font (plz note that actually I'm interested only in .otf, not in Type1), I receive the message
Bad Private Dictionary
In detail:
Elements in BlueValues/OtherBlues array are disordered<br>Elements in BlueValues/OtherBlues array are too close (Change BlueFuzz)
 Although I searched for (Google and FontForge manual) , I didn't find a clear explanation and, much less, some directions to solve that problem. I found this explanation too: https://blogs.adobe.com/CCJKType/2012/04/the-bluevalues-array.html but I dind't understan the practical way to act. Please, be kind enough to give me the basic lines to avoid all this. Thank you

Comments

  • AbrahamLee
    AbrahamLee Posts: 262
    Look at this PDF, chapter 5:

    https://www-cdf.fnal.gov/offline/PostScript/T1_SPEC.PDF
  • mauro sacchetto
    mauro sacchetto Posts: 353
    thank you. I'll read it as soon as possible.
    In any case, I tried to create a font (for merely experimental reasons) of Slimbach's Adobe Premiere Pro and I received the same warnings. So I concluded (I don't know if it's right or wrong) that it's no a so serious mistake...
  • Stefan Peev
    Stefan Peev Posts: 103
    Look at the PostScript section of Robofont Documentation. Here you will find a good decription of BlueValues.
  • mauro sacchetto
    mauro sacchetto Posts: 353
    Thank you. Reading in this moment.
    Only a question more: if I receive an error message like the ones I quoted above, can they have a negative impact on an .otf?
  • Thomas Phinney
    Thomas Phinney Posts: 2,897
    Errors in BlueValues, like those cited, can effect whether and when overshoot control turns on for small ppem sizes (primarily on screen).
  • mauro sacchetto
    mauro sacchetto Posts: 353
    ok thank, you. I've to study the case and to produce the appropriate corrections
  • mauro sacchetto
    mauro sacchetto Posts: 353
    I read the two texts, but I still do not understand the practical method for determining those values. Is there a kind of algorithm or should I deduce it by going to "measure" the glyphs?
  • Thomas Phinney
    Thomas Phinney Posts: 2,897
    There are algorithms to use to determine the interaction between BlueFuzz and the blue zones. However, the numbers depend a bit because there are several different methods for how to express the values. In FontLab Studio 5, you can use native measurements, pixels, or points. 

    BUT, the main thing to know is... you can set BlueFuzz to zero. That alone may resolve the problem. BlueFuzz is only there to catch outlines that are "off" and outside the blue zones. If the outlines are accurate, there is no need for it. The conflict arises when the BlueFuzz, added to existing zones, causes them to begin to run into each other.
  • mauro sacchetto
    mauro sacchetto Posts: 353
    edited December 2018
    I'm using FontForge n Linux (good but not perfect: I discovered a bug only a few days ago),
    but I'll try the evaluation version of FontLab in Windows.
    In any case... it seems to me that there are some fixed parameters, but in certain way it's also relative. Science and art...

Categories