poor/ugly font rendering on X11 by ghostview

Bug #59891 reported by Manuel López-Ibáñez
4
Affects Status Importance Assigned to Milestone
Debian
Invalid
Undecided
Unassigned
Mandriva
Unknown
Medium
ghostscript (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: gs

Text displayed by ghostview is poorly rendered: some parts of the characters are too light gray, they even seem missing. The problem disappears gradually as you
zoom in the document, however, even at a high zoom level, it is still noticeable.

This affects both gv and kghostview.

It occurs with both gs-gpl 8.15 and gs-esp 8.15.2

It worked fine with previous versions of Ubuntu. I had noticed that starting with version 8.X, gs has this problem in Mandriva and now in Ubuntu.

Revision history for this message
Manuel López-Ibáñez (manuellopezibanez) wrote :
Revision history for this message
Manuel López-Ibáñez (manuellopezibanez) wrote :

wrong bug watch

Changed in espgs:
status: Unconfirmed → Rejected
Revision history for this message
Manuel López-Ibáñez (manuellopezibanez) wrote :

upstream bug: http://www.cups.org/espgs/str.php?L1363

It is fine if upstream doesn't have resources or time to fix the bug right now, but at least, they should acknowledge that it exists. The attitude of Novell closing it as WONTFIX is rather surprising to me. (Another reason to prefer Ubuntu).

Revision history for this message
Manuel López-Ibáñez (manuellopezibanez) wrote :

By the way, using -dAlignToPixels=0 improves the output noticeably.

Christian Reis (kiko)
Changed in espgs:
importance: Undecided → Unknown
status: Rejected → Unknown
Changed in espgs:
status: Unknown → Rejected
Revision history for this message
Carey Underwood (cwillu) wrote :

Can you attach a ps file that shows this behaviour? The files I have handy all seem to render fine on my machine.

Does evince show the same behaviour?

Revision history for this message
Carey Underwood (cwillu) wrote :

Ignore previous comment; over the years, what you're commented has started to look normal, but I do see what you're talking about now.

Revision history for this message
Carey Underwood (cwillu) wrote :

evince (gs looks the same) vs:

gs -sDEVICE=pnm -dNOPAUSE -r600 -sOutputFile=test%d.pnm manual.ps -c quit

Much lighter, but more consistent, and as such much more readable. Whether this is useful is another matter

Revision history for this message
Manuel López-Ibáñez (manuellopezibanez) wrote :

Hum, I didn't try the PNM device but now I see that that is the work-around recommended by upstream.

http://www.cups.org/espgs/str.php?L1363

Are we going to make this the default?

Revision history for this message
Micah Cowan (micahcowan) wrote :

1. Isn't the "better" look of the pnm output only better because it was rendered at very high resolution, and so it is using the pnm viewer's antialiasing, rather than ghostscripts?
2. Does that look appreciably better than using manu's suggested -dAlignToPixels (which does indeed appear to give much better results)?

Generating a .pnm intermediary, instead of rendering properly when rendering directly, seems the wrong solution, to me.

Revision history for this message
Micah Cowan (micahcowan) wrote :

A related question: would setting -dAlignToPixels 0 as the default be a reasonable solution for this bug?

Changed in gs-gpl:
assignee: nobody → micahcowan
status: Unconfirmed → Needs Info
Revision history for this message
Manuel López-Ibáñez (manuellopezibanez) wrote :

Who are you asking that question? Even if it looks good, it may be the wrong solution. Setting -dAlignToPixels=0 seems a temporary work-around. I don't have the knowledge to judge such decision.

Micah Cowan (micahcowan)
Changed in gs-gpl:
assignee: micahcowan → nobody
status: Needs Info → Confirmed
Revision history for this message
Micah Cowan (micahcowan) wrote :

I'm asking anyone with the knowledge to answer. :-) It seems to me, though, that it's likely the right solution (certainly, more right than using pnm as an intermediary). I believe, though, that the ugly rendering may well be a result of the AlignToPixels feature working exactly as intended. That's basically what I'm asking.

Changed in espgs:
status: Invalid → Won't Fix
Changed in mandriva:
status: Invalid → Unknown
Changed in espgs:
importance: Unknown → Medium
Changed in mandriva:
importance: Unknown → Medium
Revision history for this message
Till Kamppeter (till-kamppeter) wrote :

Old bug, closing ...

Changed in ghostscript (Ubuntu):
status: Confirmed → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Bug attachments

Remote bug watches

Bug watches keep track of this bug in other bug trackers.