[osg-users] osgText inconsistant size

[Romain Ouabdelkader]

Hi Robert,

Thanks for the reply, modifying the Font*::getKerning() seems to be the
best solution.

I will cherry-pick your commit until we can upgrade to the lastest version.

Regards,
Romain Ouabdelkader.

2016-01-12 18:36 GMT+01:00 Robert Osfield <robert.osfield at gmail.com>:

> Hi Romain,
>
> Thanks for investigating and characterising this bug.  With your
> results in mind I've just done a code review of FreeTypeFont.cpp and
> Font.cpp and see that design/implementation both assume that kerning
> is invariant of font resolution, which you have found it isn't, so
> this is most definitely a flaw.
>
> From my current review the correct way to solve this would seem to be
> modify the Font::getKerning(..) and
> FontImplementation/FreeTypeFont::getKerning(..) methods to take a
> FontResolulution parameter in the same way that the getGlyph(..)
> method does, this would allow the kerning implementation to reset the
> font resolution if required as it does during the getGlyph().
>
> Unfortunately such a change would break the API so isn't something we
> can back port to OSG-3.2 or OSG-3.4, but is something we could merge
> with OSG-master/OSG-3.5.x dev series.
>
> Thoughts?
> Robert.
>
> On 12 January 2016 at 16:44, Romain Ouabdelkader
> <romain.ouabdelkader at gmail.com> wrote:
> > Hi,
> >
> > I have tracked down the issue, the difference in size come from the
> kerning
> > being calculated in different font resolutions.
> >
> > If we trace the execution:
> >
> >   osg::ref_ptr<osgText::Text> text0 = new osgText::Text;
> >   text0->setFont(font_path);
> >   text0->setFontResolution(128, 128);
> >   text0->setText("V");
> >   text0->setText("a");
> >
> > Here we generate the glyph for char 'V' and 'a' with the resolution (128,
> > 128).
> >
> >   osg::ref_ptr<osgText::Text> text1 = new osgText::Text;
> >   text1->setFont(font_path);
> >   text1->setText("p");
> >
> > We generate a dummy glyph that set FreeType face resolution to (32, 32)
> >
> >   text0->setText("Va");
> >   float first_call = text0->getBoundingBox().xMax();
> >
> > (1) Since the glyph for char 'V' and 'a' are already generated for
> > resolution (128, 128), we use the glyph in the cache (see Font::GetGlyph)
> > *without modifying FreeType face resolution*.
> > We also get kerning from FreeType but we *don't* change the face
> resolution,
> > thus the kerning is calculated in (32, 32) resolution.
> >
> >   text0->setText("c");
> >
> > We generate a dummy glyph (that isn't in the cache) that set FreeType
> face
> > resolution to (128, 128).
> >
> >   text0->setText("Va");
> >   float second_call = text0->getBoundingBox().xMax();
> >
> > Same as (1) but FreeType face resolution is (128, 128) while calculating
> the
> > kerning.
> >
> >
> >
> > This isn't a big issue, but it breaks our tests depending of the order of
> > execution...
> >
> > The issue can be resolved if Font::getKerning ask for font resolution,
> or if
> > we set font resolution at the start of Text::computeGlyphRepresentation
> (not
> > exposed in Font).
> >
> > What do you think ?
> >
> > Regards,
> > Romain Ouabdelkader.
> >
> >
> > 2016-01-06 19:16 GMT+01:00 Romain Ouabdelkader
> > <romain.ouabdelkader at gmail.com>:
> >>
> >> Hi,
> >>
> >> I have discovered an issue with osgText: under certain conditions,
> >> Text::getBoundingBox() returns different values with the same text,
> font,
> >> font resolution, etc.
> >>
> >> You can reproduce the issue with this example:
> >>
> >> int main() {
> >>   const char *font_path = "museo500.ttf";
> >>
> >>   osg::ref_ptr<osgText::Text> text0 = new osgText::Text;
> >>   text0->setFont(font_path);
> >>   text0->setFontResolution(128, 128);
> >>   text0->setText("V");
> >>   text0->setText("a");
> >>
> >>   osg::ref_ptr<osgText::Text> text1 = new osgText::Text;
> >>   text1->setFont(font_path);
> >>   text1->setText("p");
> >>
> >>   text0->setText("Va");
> >>   float first_call = text0->getBoundingBox().xMax();
> >>
> >>   text0->setText("c");
> >>
> >>   text0->setText("Va");
> >>   float second_call = text0->getBoundingBox().xMax();
> >>
> >>   std::cout << first_call << std::endl;
> >>   std::cout << second_call << std::endl;
> >>   if (first_call != second_call) {
> >>     std::cout << "Error!" << std::endl;
> >>     return 1;
> >>   }
> >>
> >>   return 0;
> >> }
> >>
> >> Output:
> >> 36.4961
> >> 35.9961
> >> Error!
> >>
> >> The issue seems to be very specific, if I remove any of these lines the
> >> issue doesn't appear.
> >>
> >> I've attached the font if you want to reproduce it.
> >>
> >> Regards,
> >> Romain Ouabdelkader.
> >>
> >
> >
> > _______________________________________________
> > osg-users mailing list
> > osg-users at lists.openscenegraph.org
> >
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
> >
> _______________________________________________
> osg-users mailing list
> osg-users at lists.openscenegraph.org
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openscenegraph.org/pipermail/osg-users-openscenegraph.org/attachments/20160113/faf508fa/attachment.htm>