A supposedly patent-free suggestion/solution to the curious subpixel rendering in Fedora

Jud Craft craftjml at gmail.com
Thu Mar 26 14:22:37 UTC 2009


No idea.  Probably something to ask the freetype people, or whoever
implements the rest of the subpixel filters.

There is someone who implemented such an idea by considering the red
and blue subpixels as dependently linked, since together red and blue
have an approximate luminance as green.  It was called the SubRGB
algorithm, and there was a page with source, a library, and examples
-- but I can't find it in google.  I don't believe it was ever used in
any font-rendering library, but it was suggested as a possible
non-patent infringing replacement.

On Thu, Mar 26, 2009 at 5:51 AM, Roberto Ragusa <mail at robertoragusa.it> wrote:
> Kevin Kofler wrote:
>>
>> Apart from the bytecode interpreter, freetype-freeworld just enables the
>> subpixel rendering in the upstream freetype. So does Fedora's freetype
>> since Fedora 10. (It was disabled in F9, it got reenabled in F10, I have no
>> idea what the rationale was for each of those decisions.) So these days the
>> only difference between freetype and freetype-freeworld is the bytecode
>> interpreter.
>
> Rendering with BCI enabled always looked awful to me, the BC-equipped fonts
> really turn into a b&w bitmap, which appears ugly to me. Maybe people who
> use Windows really like them, but my eyes are much more happy with light
> hinting and don't care about the color fringes; this could depend on the
> monitor DPI (my screen is 133dpi, I suppose 1024x768 17" laptops are
> another story).
>
> I just use the freetype from Fedora 10 and I consider my fonts beautifully
> rendered. Probably the autohinter is doing its work well.
>
> I think that font rendering on Mac is worse than on my Fedora. No colors,
> but really blurry chars; they mitigate the effect by using huge font sizes
> in the GUI.
>
> Just a thought: do I remember correctly that the color filter is just a
> simple y(0)=(x(-1)+x(0)+x(1))/3 on the subpixels? Did anyone consider that
> R, G and B have different brightness to the human eye and should be weighted
> more smartly?
>
> --
>   Roberto Ragusa    mail at robertoragusa.it
>
> --
> fedora-devel-list mailing list
> fedora-devel-list at redhat.com
> https://www.redhat.com/mailman/listinfo/fedora-devel-list
>




More information about the fedora-devel-list mailing list