Font rendering (was: Re: Flash player no longer supported, what happens to media9 and Skim?)

Jim Diamond Jim.Diamond at
Tue Aug 4 01:07:06 CEST 2020

On Mon, Aug  3, 2020 at 11:31 (-0700), Shreevatsa R wrote:

> To the original poster (Jim Diamond): could you carry out the same
> comparison (acroread versus evince) with the attached example-2.pdf
> ? Just curious about the results.

Attached find screen captures from evince and acroread, and captures
of the xmag windows.  To my eye they look the same as before, although
I didn't do a bit-by-bit comparison.

[[ Bonus attachments: b/w xmag screenshots ]]

> This example-2.pdf was generated from the given example.pdf (also re-attached, just for completeness) by running:

> ps2pdf -dPDFSETTINGS=/prepress -dAutoRotatePages=/None -dEmbedAllFonts=true -dSubsetFonts=true example.pdf example-2.pdf

That's an impression feat of compression.  woof!

> as I learned while writing an
> answer<> in 2016: the
> intended difference is that the resulting example-2.pdf uses Type 1C
> (CFF) fonts,

That article is interesting.  As are ones it points to.  And so on
down the rabbit hole.

> for which FreeType (used by Evince) has (or had) a better
> rasterizer, instead of Type 1 fonts (as example.pdf does).  All this
> may no longer hold now in 2020, but I'm just curious whether it
> makes a difference.

You can take a look at the xmag outputs and see if you see any
difference between the ones I sent before and the ones I attach here.
I'm pretty sure they are the same.

Evince's black text on white background version is certainly much more
legible than the white on blue.  From the bonus attachments you can
see that evince's output is still a fair bit different than
acroread's, but for casual reading, a far better approximation.


> On Mon, 3 Aug 2020 at 09:56, Jim Diamond via texhax <texhax at<mailto:texhax at>> wrote:
> On Mon, Aug  3, 2020 at 00:34 (-0700), Alan E. Davis wrote:

>> Off topic

> I wouldn't say so.

>> and out of order.  Please feel free to ignore this if it is
>> completely bonkers.

>> I have been pulling my hair out about differences between pdfs
>> produced by conversions by ps2pdf, of PostScript output by the Gri
>> graphing language.

> Well, OK, maybe it is off-topic.

>> This did not involved text.  A friend converted on a Mac and got
>> much prettier results than my results from ps2pdf, either standalone
>> or by way of Inkscape.  My workflow has been working fine for at
>> least two years, and other edits with Inkscape too that suddenly are
>> not working the same.

>> Why I thought this message thread relevant to my issue: I never
>> understood previous to the past month that PDFs are rendered
>> differently by different software.

> In theory, I don't think they should be.  In practice, keep in mind
> that the mathematical shapes (of the characters, as well as lines and
> curves drawn for other purposes) have to be mapped to a discrete grid
> of points.  And, as I understand it, that is where things get ugly.
> Different rendering engines can do different things.  In some cases
> you might not see the difference, but in some cases you might.

>> Even a push or a shove in some direction would be gratefully
>> received, toward learning more about this entire process.  It
>> troubles me that my work, as trivial as it is---tide graph
>> calendars, and other graphs---would not be rendered consistently on
>> different media.  I get it that in print, my carefully edited grid
>> line widths will look differently on photo paper in a consumer
>> printer than on, say, generic printer paper.

>> I have been looking at them with evince, okular, acroread (on
>> GNU/Linux), and other pdf readers.  My eye is obviously not as
>> discerning as yours, because I seem little difference between the
>> two images in your attachment.  I don't know what to look for.

> Did you read my comment that, for example, the '0' in "10 pt" looks
> different to me?  I find the evince rendering far less readable.  For
> the bigger fonts, I notice the difference between the Acroread and
> the evince rendering, but I find them both quite readable.

>> I apologize that this is probably not related to your discussion.

> I think it is, indirectly.  The reality is that if we depend on PDF as
> a document format we can share with others, it is important for other
> people to see your documents how you see them (or as closely matching
> as possible).

>> I can provide graphics if anyone wants to look at them.

> Just out of curiosity, what program did your mac friend use to convert
> PS to PDF?  (I'd be interested in some examples, but probably all I
> can do is commiserate.)

> Jim

Dr. Jim Diamond       "Convenio ergo sum"      |                        /"\
Jodrey School of Computer Science              | ASCII Ribbon Campaign  \ /
Acadia University, Wolfville NS Canada B4P 2R6 |    x
Voice: (902) 585-1402    Fax: (902) 585-1067   |                        / \
-------------- next part --------------
A non-text attachment was scrubbed...
Name: evince2.png
Type: image/png
Size: 8340 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: acroread2.png
Type: image/png
Size: 8132 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: evince2-xmag.png
Type: image/png
Size: 5582 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: acroread2-xmag.png
Type: image/png
Size: 8760 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: evince-bw-xmag.png
Type: image/png
Size: 3072 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: acroread-bw-xmag.png
Type: image/png
Size: 2921 bytes
Desc: not available
URL: <>

More information about the texhax mailing list.