Autokern from master, thoughts

Post Reply
User avatar
SCarewe
Posts: 73
Joined: 23 Apr 2021

Autokern from master, thoughts

Post by SCarewe »

Hello,

First off, many thanks for the great Autokern from Master feature. I am aware that this is the first version with this feature, but I would still like to suggest some things that have bugged me.

When setting a master to be autokerned from another source, Kern On deletes all models done up until then. This means that one cannot revert to the manually made models after checking out how it would kern from another source.

Similarly, it appears that Kern On still takes into account independent pairs set in the current master. While I understand this (even though I personally wouldn't opt for this behaviour), I am suprised Kern On doesn't allow editing or removal of the inpendent pairs. So, if I have the master set to autokern from another source, I cannot touch the independent pairs anymore, which is quite annoying.

Finally, a feature request: Would it be somehow feasible to add a tightness option for the overall kerning when kerning from another source? For instance, I have modelled a roman file, and want to use this file as the autokern reference for my italic file. My italic is spaced slightly tighter though. Kern On is now adding 10 units everywhere. I would find it very useful to have an option telling Kern On by how much the spacing is adjusted in the new master in regards to the reference. I hope this makes sense.

As always, thanks immensely for this tool (which, I hope, will spawn a spacing assistant of the same manner one day). I look forward to hearing your thoughts.
User avatar
Tim Ahrens
Site Admin
Posts: 284
Joined: 11 Jul 2019

Re: Autokern from master, thoughts

Post by Tim Ahrens »

Thanks for your comments, Sebastian!

Here are some answers:
SCarewe wrote: 12 Apr 2022 When setting a master to be autokerned from another source, Kern On deletes all models done up until then. This means that one cannot revert to the manually made models after checking out how it would kern from another source.
This action is not undoable. If you want to compare the results to a previously model-kerned version you need to work with multiple .glyphs files: Just create a backup/copy of the .glyphs file, then you can easily compare and revert as necessary.

To me, this is a very natural way to work but it seems some users dislike the idea of working with multiple/temporary files. Sorry, you will have to live with that, I am afraid.
SCarewe wrote: 12 Apr 2022 Similarly, it appears that Kern On still takes into account independent pairs set in the current master. While I understand this (even though I personally wouldn't opt for this behaviour), I am suprised Kern On doesn't allow editing or removal of the inpendent pairs.
I just fixed this in today’s update. Now you can set and un-set independent pairs.
SCarewe wrote: 12 Apr 2022 Finally, a feature request: Would it be somehow feasible to add a tightness option for the overall kerning when kerning from another source?
Yes, this is definitely something I have in mind. Something like a per-master special spacing, so to speak. Maybe that will even allow to autokern the bold from the regular, let’s see.
User avatar
Tim Ahrens
Site Admin
Posts: 284
Joined: 11 Jul 2019

Re: Autokern from master, thoughts

Post by Tim Ahrens »

What I am not sure about yet is whether and how to allow for additional models in the autokerned master. It could be frustrating if you find out that “autokern from master” generally works for you but there is a handful of pairs that get wrong kerning. What do do in this case?

(1) Strictly no additional models – one model master and one autokerned master. Instead, KO could provide a mechanism that helps you find the model you have to set in the model master so as to achieve the desired autokerning value for a particular pair in the autokerned master. This could be a bit tedious but at least it keeps things clear, no intermingling of cause and effect, so to speak.

(2) Allow for additional “local” models in the autokerned master. This is the pragmatic solution and I guess it would work. However, what to do if there are any contradictions and a model has to be removed? Maybe KO would always remove the model from the autokerned master in case of mismatches. Keep in mind that the model master may be in a different font that was modified while KO was sleeping for the autokerned master/font, and realises the mismatch only on start-up. Or, because of newly added models in the model master, some local models may become unnecessary and redundant. The whole thing may become a bit dirty unless the user is very disciplined.

(3) No distinction between the two masters/fonts, they are simply cross-linked, both can have models, and all models together autokern both masters. Seems like a cool feature but I am not sure whether our human minds are made for that. Model contradictions or inconsistency warnings would be come truly complicated in terms of UI, it would have to say something like “AT seems tighter than LT in that other font”. And, you couldn’t simply click the “show me” arrow if they are in different fonts.

Any thoughts? Thanks!
Post Reply