[cairo] UserFont size issue (ImageSurface)
behdad at behdad.org
Wed Feb 4 13:16:18 PST 2009
Ian Britten wrote:
> Behdad Esfahbod wrote:
>> You'd just need to draw in font space. That is, your drawings are typically
>> in a box of about 0 to 1 (assuming your font has zero descent).
> And if that assumption isn't true? ... :(
> It's very typical that these fonts are centred at (0,0), and thus their
> EM runs from (-x,-y), (+x,+y). Normally our driver/face hides that, but
> maybe not in this case?
Then just do a translate or something.
> > Anyway, try
>> adding a few strokes to see what's going on. Everything should just work.
> I did one better (I think), and printed out the points from glyph_path()
> (It's an 'N', upside down due to origin flipping)
> With a PDF surface, I get these points:
> move -0.3466627455219995 0.4999997600928846
> line -0.3466627455219995 -0.4999997600928846
> line -0.2533088055215585 -0.4999997600928846
> line 0.2533088055215585 0.4999997600928846
> line 0.3466627455219995 0.4999997600928846
> line 0.3466627455219995 -0.4999997600928846
> With a PNG surface, I get these points:
> move -1.2316120524949323 1.7787098309556870
> line -1.2316120524949323 -1.7787098309556870
> line -0.9000323108711681 -1.7787098309556870
> line 0.9000323108711681 1.7787098309556870
> line 1.2316120524949323 1.7787098309556870
> line 1.2316120524949323 -1.7787098309556870
> Strangely, in this case I'm now using for testing, the PNG glyphs seem
> to be drawing much *bigger* ?? Grrr - There's some sort of interaction
> going on here that I just don't see...
>> Somehow you are doing a scale somewhere maybe that you shouldn't do?
> From what I understand, and from the looks of the points, I guess I
> agree. The two sets of points should have been the same, shouldn't
> they? Same glyph, same user size, drawn in font space... The only
> apparent difference is the context they're being rendered to.
> Note that for both of these cases, the glyph was rendered at exactly
> the same user size (0.014)
> My drawing code is basically:
> Cairo::Matrix fMatrix(Cairo::scaling_matrix(userSize, -userSize));
> fMatrix.rotate(inAngle); // zero in this case
> context.set_source_rgba(r, g, b, a);
> Cairo::RefPtr<Cairo::FontFace> fFace = // make custom user font
> Here's hoping that something I've written will point to where the
> root of the problem is, or at least where to look further...
> As always, thanks for your help!
It's all in your code. Standard debugging techniques apply. Print the font
matrix and the ctm and compare. Are they different? Then check out why you
are getting different points for the two cases. The points are coming from
your code. Can't help with that.
More information about the cairo