[Pixman] font rendering

Søren Sandmann sandmann at cs.au.dk
Thu Mar 7 08:52:26 PST 2013

Gerd Hoffmann <kraxel at redhat.com> writes:

> Is there documentation somewhere describing what the pixman ops exactly do?

Unfortunately, pixman doesn't have as much documentation as it should.

> Background:  I'm trying to figure the best way to do font rendering with
> pixman.  Use case is the qemu vga text mode emulation.  I have 4 pixman
> images involved:
>   (1) The (virtual) vga framebuffer, in PIXMAN_x8r8g8b8
>   (2) The vgafont glyph, in PIXMAN_a8r8g8b8,
>       with 0xffffffff / 0x00000000 for font/background pixels.
>   (3) The foreground (font) color, created using
>       pixman_image_create_solid_fill()
>   (4) The background color, likewise created using
>       pixman_image_create_solid_fill()
> What I'm trying now is this:
>     pixman_image_composite(PIXMAN_OP_SRC, bg, NULL, vgafb, ...);
>     pixman_image_composite(PIXMAN_OP_OVER, fg, glyph, vgafb, ...);
> So the first op fills with the background color, and the second renders
> the character.  Comes close to what I want.  Except that the second
> blends the foreground color with the background color instead of
> replacing it.

If I understand correctly, what you are doing now should work. Blending
of the foreground and background colors should only happen if the
foreground color has alpha != 0xff, or if a glyph pixel has alpha that
is neither 0x00 nor 0xff. In either case, blending is presumably what
you want?


More information about the Pixman mailing list