Bug 691629

Summary:	Mupdf: Rendering artifacts with mispositioning of scaled images
Product:	MuPDF	Reporter:	av1474
Component:	fitz	Assignee:	Robin Watts <robin.watts>
Status:	RESOLVED FIXED
Severity:	normal	CC:	jackie.rosen, mehmetgelisin, robin.watts, zeniko
Priority:	P4
Version:	unspecified
Hardware:	Other
OS:	Linux
Customer:		Word Size:	---
Attachments:	Sample file.

Description av1474 2010-09-19 10:49:12 UTC

http://csseminar.kadm.usu.ru/tarragona_volkov2008.pdf is mis-rendered with
mupdf's GIT HEAD (and ghostscript 8.71/9.0 too) however version of MuPDF from
SumatraPDF's repository is OK.

Comment 1 Alex Cherepanov 2010-09-19 14:50:34 UTC

Created attachment 6738 [details]
Sample file.

This is a local copy of the sample file.

The file is drawn as a sequence of black and white images drawn on the black
background. Sometimes, there's a one-pixel positioning difference between the
image and the background that results in a black line across the page.
On Ghostscript, the problem disappears at the resolution that is a
multiple of 200 dpi.

Comment 2 Robin Watts 2011-07-27 22:59:33 UTC

Taking this, as it's to do with scaling.

Comment 3 Robin Watts 2011-08-01 11:54:14 UTC

This problem is to do with images having antialiased edges, even after gridfitting.

The weights at the edges of images are slightly less than full, and this is most easily rectified by wrapping the weights slightly.

Done in my local repo. Will close bug when the commit makes it to the trunk.

Comment 4 zeniko 2011-08-01 19:29:31 UTC

FYI: http://git.ghostscript.com/?p=mupdf.git;a=commitdiff;h=d451b09f75625a1b902c14afdede5242ad66713a significantly regresses the rendering of Type 3 fonts for which the glyphs are stored as images (see e.g. the document attached to http://code.google.com/p/sumatrapdf/issues/detail?id=1046 ).

Comment 5 Robin Watts 2011-08-02 14:03:28 UTC

Fixed in commit:

commit b076818f88a3ab152897cd2e65c95cef9afb471f
Author: Robin Watts <Robin.Watts@artifex.com>
Date:   Tue Aug 2 14:06:18 2011 +0100

    Another attempt to fix bug 691629.

    To solve bug 691629 we need to ensure that the scaling weights for
    every pixel in a gridfitted image sum to 256. I had attempted to
    do that by enabling the 'WRAP' code, but this has the effect of
    adding too much bias to the outlying source pixels, resulting in
    thickened serifs etc.

    A better fix is to extend the code that is already present to check
    the weights for validity. If an image pixel is completely covered,
    then force the weights to 256 by adjusting the largest weight.

    This still skews the output slightly but it's a much less visible
    result.