FL TransType 4

2

Comments

  • Hello James,

    I've heard about this, it's rather strange. Does this happen when you drag-and-drop as well as when you use File / Add Font Files or File / Add Folder?
  • The user and all related content has been deleted.
  • The user and all related content has been deleted.
  • James, does your path include any special characters, such as TM? I had a similar problem and that proved to be the culprit.
  • The user and all related content has been deleted.
  • The user and all related content has been deleted.
  • I must say it the interface look fresh and the conversion was very quick.
  • We've just published the 4th beta version of TransType 4 for Mac OS X. Set to expire on March 31st, this beta brings some small improvements to the UI, including an expanded Credits and Licensing section that allows you to view the EULA embedded in the font. More changes and improvements, and a Windows version, to come soon! You can download TransType 4 from http://www.fontlab.com/
  • Next public beta of TransType 4 for Mac is here: http://bit.ly/trt4macbeta :)
  • James PuckettJames Puckett Posts: 1,969
    edited June 2013
    I tested the latest beta and it makes a good web font converter. The rest of the functionality is pretty much a joke in the year 2013. TransType is definitely faster and easier to use than FontPrep. I might buy this as a $20 web font conversion tool to save me from messing around with Fontforge But only if I could disable the annoying habit TransType has of changing my file names. I would pay twice that if it would hook into my installed version of TTFA and save me from having to run multiple programs to do quick tests of new fonts.
  • Adam TwardochAdam Twardoch Posts: 507
    edited June 2013
    James,

    many thanks for your feedback! All things that you mention are planned for an edition of the product that will be targeted more specifically at the font foundries.

    One thing I can say, though, is that the by far most common problem reported by many people has always been the difficulty to get the family naming right, including style linking etc. — and it still continues to be the case in 2013.

    During TransType testing, we've uncovered a remarkable number of naming errors in larger font families (in shipping fonts made by different renowned vendors), where the PostScript name, the weight information or the copyright string was "off" in one of the fonts. Sometimes, this is leading to problems on the users' end. We've designed the TransType user interface to address this.

    Everyone has somewhat different requirements for what a "technical font production tool" should do, and our goal is to address the most common ones in a product such as TransType 4, and give more demanding customers a more advanced tool which offers them more flexibility. Supporting different kinds of file naming schemes is among the more advanced functionalities.

    Best,
    Adam
  • Adam TwardochAdam Twardoch Posts: 507
    edited June 2013
    Ps. James,

    As for ttfautohint, we've been delaying the integration of it into our tools due to ttfautohint's well-known problems where the hinting generated by ttfautohint triggers a behavior in some rasterizers that produces the "exploding glyphs" effect. This happens on a number of popular printers, e.g. HP LaserJet 4350. We feel that as long as ttfautohint produces TrueType hinting code that causes printouts like the one shown below, it would be premature for us to integrate it — at least without making sure that users are properly informed about the potential caveats.

    Or would you prefer if we integrated ttfautohint regardless of such problems?

    Thanks!
    Adam
  • As you're here with the news Adam, do you have any news about FLS 6?
  • James PuckettJames Puckett Posts: 1,969
    Supporting different kinds of file naming schemes is among the more advanced functionalities.
    I don’t really need anything special. I just need to be able to turn it off so that NewFont.beta1.ttf doesn’t turn into four files names NewFont.ttf that I need to rename.

    I was not aware of that issue with TTFA. I can see why it would scare you off!
  • Thomas PhinneyThomas Phinney Posts: 2,730
    Even though I am aware of the TTFA issue, and it bothers me a lot, for web font usage, I am willing to take that tradeoff versus using another autohinter with worse results on screen.

    So I for one would welcome TTFA integration for my needs, although I can understand why it FontLab might choose to avoid it for now.

    T
  • What is the root cause of the sposions in TTFA? Is it repeatable or does this happen randomly?
  • What is the root cause of the sposions in TTFA? Is it repeatable or does this happen randomly?
    Could it be crappy Bezier drawings and even crappier automatic TTF outline conversion?
  • Deleted AccountDeleted Account Posts: 739
    edited June 2013
    "Or would you prefer if we integrated ttfautohint regardless of such problems? "

    'Splodin' printouts on a 1200 dpi printer, are stopping the mob from autohinting for the web !@#$%? Run.

    It's 1 line of tt per font to turn off hinting > any given ppm, and it's just set way too high in ttfautohint... if that's a hap'nin on a 1200 dpi printer brother Adam.

    I would consult the oracle of Vienna.
  • Thomas PhinneyThomas Phinney Posts: 2,730
    edited June 2013
    Excellent point from DB. But we believe this affects some 600 dpi devices as well. I suppose we could turn off hinting at 50 ppem and above, leaving 6 pt type at 600 dpi unhinted and "safe" from the bug. Not ideal, but probably better than the alternatives, eh wot?
  • David, that's an excellent point. I just received information from another user that the same problem happens with Canon MP550. I'll do some experiments with turning off instructions above a certain limit.

    The problem is in the implementation of the division operator in some rasterizers. While the TT spec prescribes that division should truncate (so 5/3 should be equal 1 in the TrueType hinting language), some implementations perform rounding (so 5/3 = 2). Ttfautohint uses division extensively, much more often than other hinting tools, and it does this repeatedly, so the errors accumulate, resulting in point coordinates being "blown up".

    I'm corresponding with the people involved so that ttfautohint can be patched. One possible solution is for the ttfautohint-generated bytecode to do a "test division" (e.g. 5/3) and compare the result with an expected value. If the value is 2, then hinting could be disabled, and if the value is 1, hinting would be executed. This may take some time to implement, but — since we at FontLab do want to integrate ttfautohint (after well, we donated money to its development) — I'm trying to find a workable solution.
  • Deleted AccountDeleted Account Posts: 739
    edited June 2013
    I fully agree, if you funded it, you should get something groovy to work.

    But how does 1.4 (5/3) round to two again? What TT interpreter?
  • John HudsonJohn Hudson Posts: 2,955
    But how does 1.4 (5/3) round to two again?
    By being 1.6666... not 1.4
  • John HudsonJohn Hudson Posts: 2,955
    One possible solution is for the ttfautohint-generated bytecode to do a "test division"
    The other possible solution is to blame the buggy rasterisers that are not following the TrueType specification. Why the heck should ttfautohint be made to jump through hoops and fonts have to disable functionality to accommodate other peoples' obvious mistakes? Why the heck even have a specification if font makers and font tool makers are just going to bend over backwards to allow people to ignore it?
  • Thomas PhinneyThomas Phinney Posts: 2,730
    We can blame the rasterizer all we want, but for many customers, all they will care about is that our fonts screw up when printing on their incredibly common printers.

    In this case, the rasterizer is being (or has been) fixed, but because it is embedded in hardware devices, it will be a problem for a Very Long Time. :(
  • John HudsonJohn Hudson Posts: 2,955
    Oh, I know, I know ... but it sure rubs me the wrong way that decade after decade this sort of thing persists because doing the wrong thing gets rewarded.
  • All we can do, (besides using calculators), is make a special font for every single purpose, device, size, language, and modem type.
  • Chris LozosChris Lozos Posts: 1,458
    and dog sled.
  • I meant no disrespect to iditarodic users.
  • Chris LozosChris Lozos Posts: 1,458
    Now ya made me get all mushy, David.
  • Adam TwardochAdam Twardoch Posts: 507
    edited June 2013
    Author of ttfautohint Werner Lemberg reports that he seems to have fixed the problem, i.e. change the TT hinting code the tool generates to that it works around the rasterizer bug. I was not able to verify it yet, but the improved code is said to be in the ttfautohint git repo. If anyone could build a Mac OS X commandline version of the current source (as I'm not able to do it on my current system) and relay it to me, I could test it and report my findings.
Sign In or Register to comment.