search

Font-Kai / Kaisei-Opti

6 stars 1 forks source link

Latin does not have a consistent baseline #2

Closed aaronbell closed 3 years ago

aaronbell commented 4 years ago

Between the W4, W5, and W7 weights, the Latin does not have a consistent baseline.

Screen Shot 2020-11-02 at 7 33 11 PM

While this may work in a Japanese-first model, fonts on Google Fonts need to be able to be used for Latin as well, and this sort of baseline variation will result in an unexpected behavior should a Latin user change weights. It would be better to follow the model used by Noto JP wherein all the weights are on the same baseline, and then a BASE table is used to vary the baseline as necessary.

The BASE table can be inserted into Glyphs in the featues tab: https://forum.glyphsapp.com/t/name-table-entry-custom-parameter/2257

aaronbell commented 4 years ago

To expand on this, I made a Glyphs file that contains each of the four weights W4 W5 W7 and W8 from across the Font Kai library, and am observing a lot of inconsistencies in placement / weighting.

For example, in the dagger.rotat:

Screen Shot 2020-11-03 at 9 43 44 AM

The W8 version is heavier, but centered in the BBOX, whereas the other three are at the top of the box.

Or partialdiff, which is pretty consistently weighted until the W8 version:

Screen Shot 2020-11-03 at 9 45 36 AM

In zero-han, the W5 looks to be the heaviest, followed by W7, then W4 then W8:

Screen Shot 2020-11-03 at 9 46 39 AM

Or the ordmasculine, where it changes proportion and position:

Screen Shot 2020-11-03 at 9 48 40 AM

I'd love to see these ironed out and generally made more consistent from lightest to darkest. For your convenience, I have uploaded a zip file that contains the Glyphs file with shared W4–W8 weights for your consideration here: https://github.com/aaronbell/Kaisei-Opti/tree/shared_glyphs/sources