Behdad Esfahbod
Behdad Esfahbod
Fixed in https://github.com/fonttools/fonttools/pull/3506
> Interestingly, harfbuzz instancing produces the same artifacts, but AFDKO’s tx does not. @qxliu76
Random thing: if the test is CFF2, why does it have `endchar` operators?
That suggest some lazy-loading is not happening?
Tangent: It really sucks that Adobe people decided to add user-space variable values to the .fea format, since those can't work with avar2.
> I'm also not quite sure what the fundamental problem is: these coords get normalized without applying any avar, and that allows us to express everything that's needed, even if...
But the fundamental issue is that with avar2, any point in the user space can be mapped to a point in the normalized space. But the same is not necessarily...
> Do you happen to know why it looks completely different? This was changed in harfbuzz in https://github.com/harfbuzz/harfbuzz/commit/a150baf32c
Very interesting and bizarre. I'll try to reproduce. Just checking, what rasterizer is that renderer using? FreeType?
> It looks like the IUP points are not rounded to the unit grid, and we get ever so slightly different rounding errors Interesting. Yes, that would explain this. I...