Re: [ft-devel] New `slight' auto-hinting mode

[Werner LEMBERG]

>> Well, lsb_delta and rsb_delta are derived from the horizontal
>> phantom points.  Given that the idea of `phantom points' is
>> originally a TrueType concept, I don't see why this shouldn't work
>> with TrueType hinting.
> 
> Currently there is a need to adjust pen position when hinting places
> two adjacent glyph edges too near or too far.  This concept does not
> go away with subpixel pen positioning.

Why do you think so?  Where is the difference to the integer case?
TrueType bytecode instructions also adjust the advance width together
with the lsb and rsb (by sometimes more than a pixel!) without looking
at the previous glyph.  For sub-pixel rendering, it would be necessary
to know the fractional position of the current glyph as an additional
constraint so that bytecode instructions can compute better grid
coverage, but again you don't have to look at the previous glyph.

> Essentially, this mean that each glyph advance *must* depend on the
> previous pen position.

Maybe we are miscommunicating.  Can you give an example for your
claim?

> I doubt that this is easier to comprehend than lsb_delta and
> rsb_delta.

I'm not talking about comprehension.  My main concern is your
statement that sub-pixel hinting along the horizontal axis and
fractional advance widths don't work together, which I consider plain
wrong.


    Werner