[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#131336: more information



reassign 131336 libt1-5
forwarded 131336 Rainer.Menzner@web.de
thanks

Stefan Ulrich <stefan-ulrich.nntp@zen.co.uk> wrote:

[Fullquote because I'm assigning to an other package, answer is below]

> Matt Swift <swift@alum.mit.edu> writes:
>
>> I believe I responded to this question on 29 June 2003 by sending a
>> tarball with two X-windows screen dumps.  I can download the tarball
>> today by visiting
>
>> http://bugs.debian.org/cgi-bin/bugreport.cgi/testdvi.tgz?bug=131336&msg=12&att=1
>
> Oops, I didn't see this followup until now ...
>
>> As I remarked, this difference may not be noticeable with another
>> Metafont mode.  At the moment I do not recall exactly why "ricoh" is a
>> better mode choice than "cx" (I think the default) for XDvi PK fonts,
>> but if "cx" looks as bad as the T1 does on my system, then neither is
>> acceptable.  
>
> Interesting. I always recommend using a 600 dpi mode since all
> modes < 600 dpi really do give suboptimal results with
> anti-aliasing, with the ratio of unshrunken bits to dots on the
> screen being too small. Also, 600dpi modes give you more shrink
> steps to choose from. (For this very reason, setting the metafont
> mode for xdvi has been removed from current texconfig, and the
> default resolution is set to 600dpi). These days, additional PK
> fonts shouldn't be an issue, especially if you're using t1lib.
>
> With 22.40y1, which uses the same t1lib version as 22.40v, the
> results look OK at 600dpi. It looks like the results with the
> 300dpi ricoh mode are the result of suboptimal hinting. The
> hinting code has changed in current t1lib (which is used by
> xdvik > 22.7x), but unfortunately the new code gives an even
> *worse* result at 600dpi; the horizontal bar of the `e' is only
> 1 pixel wide:
>
> http://xdvi.sourceforge.net/600dpi.gif
>
> Only 1200dpi (e.g. xdvi -mfmode lexmarkr:1200) gives a
> good result:
>
> http://xdvi.sourceforge.net/1200dpi.gif
>
> I'll write a bug report to Rainer Menzner, the t1lib author ...

I'm changing the forwarded-to address in the Debian BTS, assuming that
you're doing this promptly.

Matt's screenshots, taken in 2003, where probably made with Debian woody's
libt1.so.1.3.1, not with current libt1.so.5.0.0. Don't know what you,
Stefan, where using.

Regards, Frank

P.S. to the t1lib maintainers: I guess bugs.debian.org is faster in
forwarding mails to maintainers, than control@b.d.o is in changing the
maintainers' address. Therefore I'm sending you a separate copy.

-- 
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie




Reply to: