Kern bouncy letters? Figure out a way to do this without unnecessary kerns.

[arrowtype]

I initially envisioned the flux & bounce axes as temporary, fun effects that could be played with for animation, interaction, and casual usage where spacing wasn’t as important as overall the fact that it was bouncy. I assumed that the build would be best if it randomized the fonts with each build, to create more variety.

However, users will inevitably want to use the maximum bounce & flux settings for something like a logo or storefront sign, and this would freeze an instance of bad spacing due to bounce as a bad look, permanently. Moreover, some uses would be extremely disrupted if bounce results are quite different between versions – making tiny "hot fixes" basically impossible to make. At the end of the day, this is still a typeface, and it needs to be possible to use and maintain in the long term.

Therefore, it is pretty important that:

• It is possible to build fonts with repeatable results
• It is possible to tweak kerns where needed in the bouncy & flux styles
• Kerning in Bouncy sources should be preserved when new bouncy sources are generated.
• Amount of bounce should also be recorded and preserved when new bouncy sources are generated.

The bounce-generation script, then, should:

• record bounce values to each UFO’s lib.plist
• compare Y min (?) to core UFO to check if bounces have been adjusted manually, before next build

The end result is that the build workflow should allow for randomization where desired (basically during the design phase), but also allow for improvements to be made over time, rather than be overridden.

[arrowtype]

In the end, to have alternating glyphs (a, b.alt1, c.alt2, a, b.alt1, c.alt2, etc) and to keep non-bouncy glyphs kerned, all alts had to be added to the kerning groups of the base glyphs.

I did end up recording bounce values to the UFO libs, to make builds reproducible and to allow for tweaks to bouncing, but kerning in bounced styles would blow up the scope, AND it would prevent duplexed bouncing animations, which are a core design goal for this axis. So, I’m closing this issue as complete.