Re: Reasonable parameters for reproducing character rendering issues?

[Brad Neimann]

Hi Craig,

Unfortunately, it’s been a while since I last thought about this issue, 
so I don’t recall precisely how I reproduced it. My methodology was, 
basically, ‘play around with ftgrid with forced auto-hinting until I 
find a character with issues’. And I have a horrible feeling I might 
have using the wrong settings in ftgrid in any case...

That being said, I do recall that I needed to go down to very small 
sizes to find issues --- when there’s enough space FreeType is already 
pretty good at hinting! Trying again with ftgrid, it seems that 
characters ⟨ij⟩ in DejaVu Sans at 7.5 ppem are really bad: if you 
increase the font size by even 0.5 ppem more they suddenly get a lot 
better. Pushing those upper dots just one pixel higher would solve the 
issue. For comparison, at 7.5 ppem most other characters in the font are 
still quite distinct. If you have access to it, Calibri below 8.5 ppem 
also illustrates the problem: the dots on ⟨ij⟩ both nearly vanish at 
those sizes. (Incidentally, Calibri’s encoding also allows easy 
comparison between the various diacritics, and on that basis I’d say 
that ⟨ì⟩ at 8.5 ppem is a reasonable approximation to what I’d want ⟨i⟩ 
to look like.) And for an example of a font which *doesn’t* have such 
great issues, the dots on ⟨ij⟩ in Noto Sans remain fairly distinct even 
when the letters themselves are near-illegible.

Regards,
Brad