Good point - the bug could actually be in the rendering, rather than in the bounding box calculation.
We could determine which is the case, by running the exact same script with the exact same font, on a system where the bug is present, and on another system where it's not. Then overlay the resulting images in photoshop: - if the type aligns, and the red lines do not, the bug is in the bounding box computation. - if the type does not align, the bug is either in the type rendering, or in both (less likely) the rendering and bounding box computation. That should narrow it down some? On Fri, Sep 7, 2012 at 1:12 AM, Pierre Joye <pierre....@gmail.com> wrote: > On Fri, Sep 7, 2012 at 3:09 AM, Rasmus Schultz <ras...@mindplay.dk> wrote: > > Jannik, > > > > Thank you - this confirms I'm not crazy, or at least it's evidence to > > support that theory ;-) > > > > It may be OS specific - perhaps the Windows and OSX binaries are built > > against a different build (or configuration) of FreeType? > > It can depend on three things: > > - bug in gd (bbox or offset calculation) > - freetype version > - freetype options (some options, not always enabled, strongly affects > rendering) > > cheers, > -- > Pierre > > @pierrejoye | http://blog.thepimp.net | http://www.libgd.org >