Re: [ft-devel] inconsistently dropping sub-pixel features

[Werner LEMBERG]

> KS> If the feature lies completely below one half pixel (ie yMax <
> KS> 32) then the whole glyph is elided. If any part of the glyph
> KS> lies above half the pixel (ie yMax >= 32) then the pixel is
> KS> marked.
>
> Interesting.

It's easily possible that my overshoot code doesn't cover this corner
case.

> I wonder whether the Overshoot_Top and Overshoot_Bottom flags remain
> constant when the glyph is flipped with FT_Set_Transform()?

What do you mean with `constant'?  The transformation takes place
before the glyph gets rendered, and those flags are computed during
the rasterization.

> If Vertical_Sweep_Drop() is the problem, chaning it to test more
> than just >=50% may have significant effects elsewhere.

Actually, the dropout algorithm with stubs is still not working in all
cases; I simply can't comprehend the meaning of the specification, and
whatever I tried, there are some cases where the results differ from
the MS rasterizer's results.  The wording of the algorithm description
has basically been taken without modifications from Apple's ancient
TrueType specs – I suspect that it is intentionally fuzzy...


    Werner