Re: [ft-devel] RE: dropout in '2' in times.ttf

[David Turner]

Hi Graham,

as you said, the real problem is that the bytecode program for the "2"
distorts the vector so much that it creates an ultra-thin diagonal stem.

if you're not using the auto-hinter, which shouldn't have this problem,
you should either use monochrome rendering, or use a different technique.

for instance, render *and* hint at a larger resolution, then filter the
result back to the original size. Of course, one problem is that the shapes
and metrics you're going to get will be slightly different anyway.

Regards,

- David Turner
- The FreeType Project  (www.freetype.org)

On Tue, 22 May 2007 14:42:58 +0100, "Graham Asher" <address@hidden> said:
> Please ignore this: I now have a fix, which is to turn off hinting when
> anti-aliasing. It seems that the effect of grid-fitting the 2 (as I
> verified
> by looking at it in FontForge) is to so narrow the stem that it can only
> be
> rescued by dropout control; which doesn't exist for anti-aliased
> rendering.
> 
> The question still arises of how to sensibly apply hinting to
> anti-aliased
> characters, which can still benefit from snapping stem widths to integral
> pixels, particularly at small sizes.
> 
> Best regards,
> 
> Graham
> 
> 
> -----Original Message-----
> From: Graham Asher [mailto:address@hidden 
> Sent: 22 May 2007 12:57
> To: 'address@hidden'
> Subject: dropout in '2' in times.ttf
> 
> Dear FreeTypers,
> 
> Here is a problem that has probably already been solved, given the
> popularity of Times Roman.
> 
> Is there a solution for the really bad dropout in the digit '2' in Times
> Roman (times.ttf) when anti-aliased? I enclose a PNG file illustrating
> the
> problem at 16ppem.
> 
> Here are the actual pixel values generated by FreeType 2.3.4, as 11 rows
> of
> 6 pixels, in hex:
> 
> 6 93 F0 E0 43 00 
> 2 65 0A 2D CD 2F 
> 6E 00 00 00 3D C5
> 0 00 00 00 06 EB
> 0 00 00 00 1F 9A
> 00 00 00 00 6B 1B
> 0 00 00 1A 3C 00
> 00 00 06 25 00 00
> 0 02 11 00 00 00
> 4 10 00 03 4E BB
> B4 FF FF FF FF 88
> 
> Note the stem near the bottom has adjacent values 2h and 11h, totalling
> 13h
> = 19; thus the stem has an effective horizontal width of 19/255 = 0.075
> pixels, which is incredibly narrow. Naturally the pixels don't show up at
> all.
> 
> Thanks in advance for any help or advice.
> 
> Graham Asher
> 
> 
> 
> 
> _______________________________________________
> Freetype-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/freetype-devel