FreeType2: Get global font bounding box in pixels?

I'm using FreeType2 for font rendering, and I need to get a global bounding box for all fonts, so I can align them in a nice grid. I call FT_Set_Char_Size followed by extracting the global bounds using

int pixels_x = ::FT_MulFix((face->bbox.xMax - face->bbox.xMin), face->size->metrics.x_scale );
int pixels_y = ::FT_MulFix((face->bbox.yMax - face->bbOx.yMin), face->size->metrics.y_scale );
return Size (pixels_x / 64, pixels_y / 64);

which works, but it's quite a bit too large. I also tried to compute using doubles (as describe开发者_开发百科d in the FreeType2 tutorial), but the results are practically the same. Even using just face->bbox.xMax results in bounding boxes which are too wide. Am I doing the right thing, or is there simply some huge glyph in my font (Arial.ttf in this case?) Any way to check which glyph is supposedly that big?

Why not calculate the min/max from the characters that you are using in the string that you want to align? Just loop through the characters and store the maximum and minimum from the characters that you are using. You can store these values after you rendered them so you don't need to look it up every time you render the glyphs.

I have a similar problem using freetype to render a bunch of text elements that will appear in a grid. Not all of the text elements are the same size, and I need to prerender them before I know where they would be laid out. The different sizes were the biggest problem when the heights changed, such as for letters with descending portions (like "j" or "Q").

I ended up using the height that is on the face (kind of like you did with the bbox). But like you mentioned, that value was much to big. It's supposed to be the baseline to baseline distance, but it appeared to be about twice that distance. So, I took the easy way out and divided the reported height by 2 and used that as a general height value. Most likely, the height is too big because there are some characters in the font that go way high or way low.

I suppose a better way might be to loop through all the characters expected to be used, get their glyph metrics and store the largest height found. But that doesn't seem all that robust either.

Your code is right.

It's not too large.

Because there are so many special symbols that is vary large than ascii charater. . view special big symbol

it's easy to traverse all unicode charcode, to find those large symbol.

if you only need ascii, my hack method is

FT_MulFix(face_->units_per_EM, face_->size->metrics.x_scale ) >> 6

FT_MulFix(face_->units_per_EM, face_->size->metrics.y_scale ) >> 6