http://www.typophile.com/node/19646

This doesn´t work like it should, lots of people creating a full weight family should have experienced this.. no?

From having 250 kerning pairs in the MM font, down to only 89 in the generated instance, that is a significant lost. Can you tell, in general, which pairs are being dropped out?

It´s hard to say, it seems new pairs are created that I never defined, and most of those I did define are gone.. It´s totally screwed up. I make the MM font with the “blend fonts”-function, then generate an instance out of this. Is this the wrong way to do it?

> Is this the wrong way to do it?

Not necessarily. FL offers more than one way to put together a MM font; hopefully these will all produce a reliable source for generating instances.

Since the issue here is the kerning data, when you blend the fonts, do any of them have kerning? If so, do the pairs ’overlap’? (Meaning, a given pair exists in both fonts)

When you open a Metrics Window, you should be able to see the various kerning values for each master. Do they look alright?

I’m using FL 5.0.2 Mac, and from the tests I’ve done with a 2-axis (4 masters) font, I had no problems. No pairs were dropped or added, and the kerning was interpolated.

The two masters both have about the same number of kerning pairs with many having the same value, yes.. Most pairs exist in both masters, and they look totally normal to me. Maybe they fixed it in 5.0.2 then.. Thanks anyway Miguel!

I can confirm that we have made some improvements in kerning handling in FontLab Studio 5.0.2. I’m not sure if they’re MM-related but chances are that 5.0.2 will help you.

Regards,
Adam Twardoch
Fontlab Ltd.

Tried it in the .02 version too now, with the same result..strange

Got it now.. it does work in 5.0.2 when defining a new axis.. but not with the “blend fonts” function!

Give this a try: (Can’t promise it’ll work, tho.)
1. Extract the kerning of both fonts to .AFM files, before blending them;
2. Blend the fonts, and then import the kerning back from each file into the correspondent master.

It works like this: In the font where you set up the masters, you have to import the metrics/kerning data into each corresponding master - before generating an instance, otherwise all masters get the same data! It shouldn´t be like that..