RazrFalcon
In what units are GylphPositions?
After shaping, we can obtain GlyphPosition for each glyph. However, I can't figure out what coordinate unit it uses. The positions are proportional to what it supposed to be, but I don't know the actual ratio and its rationale.
Also, is there a way to set the font size before the shaping process? In original HarfBuzz, we can do it via hb_font_set_ppem (as described here). Integrated with FreeType, HarfBuzz reads the font size in FreeType font struct when hb_ft_font_create-ing. But I could not find the corresponding method in rustybuzz::Face struct. set_points_per_em did not work, as it does not change the values of GlyphPositions. (Specifically, I ran example/shape.rs, but enabling or disabling the option did not affect the output.) So how can I do that? Do I have to scale the coordinates manually according to what font size I need? In such case I need the scaling factor, as I stated at first. Or am I missing something?
Values are in font units, aka units per EM (ttf_parser::Face::units_per_em). You can get the scale factor like this: font-size-in-px / units-per-em and the multiply each value by this factor.
The is no way to set font size in the moment. I will add it in the future.
Thank you! Do we need to parse the binary font file once again to obtain ttf_parse::Face instance? Or is there a way to convert it from rustybuzz::Face?
Yes, for now you have to parse it again, but it's pretty cheap in case of ttf-parser.
Just out of curiosity, what do set_pixels_per_em and set_points_per_em do? It seems they does not change the outcome.
@laurmaedje I see, thanks for your information. @RazrFalcon I tested the ratio and it worked like a charm! Thank you.
Should I close the issue or leave it open until "set font size" feature is added?
Values are in font units, aka units per EM (
ttf_parser::Face::units_per_em). You can get the scale factor like this:font-size-in-px / units-per-emand the multiply each value by this factor.The is no way to set font size in the moment. I will add it in the future.
So if I want to typeset some text using Font A, I would have to parse the font and get the face I want, store the units_per_em in a variable, get the glyphs for my text, calculate the ratio by which I have to scale x_actual and y_actual and then multiply these values with the ratio. That should be my final position. These values are then in pixels, right?
I am having trouble to get the right values since I want to render the glyphs into a PDF. And that PDF takes points as its coordinates.
There are no pixels in fonts, since they are vector and not raster. All you can do is to scale them by whatever factor you want/need.
Let's say your font has units-per-em set to 2048. Those are points. If you want to render them 1:1 then you will need 2048x2048 pixels canvas. By scaling everything down you can fit it into whatever canvas you want.
As for PDF, it uses units-per-em as well. The only difference it that some values are in default PDF units (1000) and some are in font units (units-per-em).
As for rustybuzz and ttf-parser, they operate in font units. Meaning they do not scale anything.