Hi Tim,
These classes are ok?
Wrong class?
Wrong class?
- Attachments
-
- Screen Shot 2021-12-27 at 20.13.24.png (75.81 KiB) Viewed 4796 times
-
- Screen Shot 2021-12-27 at 20.12.57.png (104.84 KiB) Viewed 4796 times
- Tim Ahrens
- Site Admin
- Posts: 415
- Joined: 11 Jul 2019
Re: Wrong class?
Hello Eduardo, don’t worry about the class names. The only thing that matters is the final rendering. Is there any pair that is wrong in the final result (i.e. when the font is used)?
The grouping of glyphs in the classes can look a bit strange to human eyes, as it is generated with storage space in mind. If you want you can send me the file and I can have a closer look. Also, which kB size did you specify (and how many glyphs does your font have)?
The grouping of glyphs in the classes can look a bit strange to human eyes, as it is generated with storage space in mind. If you want you can send me the file and I can have a closer look. Also, which kB size did you specify (and how many glyphs does your font have)?
Re: Wrong class?
Thanks Tim,
My font have 600 glyphs [2 masters] at 24 kB. Resulting about 6500 pairs per master that include zero kerning pairs. With this configuration the class appear ok.
But I think that I have unnecessary pairs like sample and don't know how resolve it.
Let me know if you need my glyphs file.
Cheers, Edu
My font have 600 glyphs [2 masters] at 24 kB. Resulting about 6500 pairs per master that include zero kerning pairs. With this configuration the class appear ok.
But I think that I have unnecessary pairs like sample and don't know how resolve it.
Let me know if you need my glyphs file.
Cheers, Edu
- Attachments
-
- ellipsis-quotedbl
- Screen Shot 2021-12-28 at 15.52.22.png (25.39 KiB) Viewed 4789 times
- Tim Ahrens
- Site Admin
- Posts: 415
- Joined: 11 Jul 2019
Re: Wrong class?
It looks like the pair you are showing is /ellipsis/quotedblleft? I’d say it’s certainly necessary!
Re: Wrong class?
Tim Ahrens wrote:
> It looks like the pair you are showing is /ellipsis/quotedblleft? I’d say
> it’s certainly necessary!
Do you think that the number of pairs generated from my file it's ok?
> It looks like the pair you are showing is /ellipsis/quotedblleft? I’d say
> it’s certainly necessary!
Do you think that the number of pairs generated from my file it's ok?
-
- Posts: 4
- Joined: 23 Apr 2021
Re: Wrong class?
Hi Tim,
I understand that it's done with storage space in mind, but this system complicates quite a lot any manual intervention after the fact. Is there a way to achieve "human" classes? Or could it become an option in the future?
BTW the plugin rocks!
a
I understand that it's done with storage space in mind, but this system complicates quite a lot any manual intervention after the fact. Is there a way to achieve "human" classes? Or could it become an option in the future?
BTW the plugin rocks!
a
- Tim Ahrens
- Site Admin
- Posts: 415
- Joined: 11 Jul 2019
Re: Wrong class?
I don’t think the number of kerning pairs matters at all. Only the size of the kerning data is relevant.
For desktop fonts, as long as you stay below 65 kB you will not get any subtable overflow errors on export, and I don’t think the file size matters.
For webfonts, you need to decide how much kerning weight you want to allow. It’s a matter of balancing kerning accuracy and file size, i.e. how fast the website will load. The webfont users, i.e. the website makers, have very different opinions on this: Some don’t care about the webfont file size while others will fight for every kB that’s saved, even if that slightly lowers the quality of the rendered text.
Keep in mind that the size you specify in Kern On is the raw kerning weight and it is compressed as you create a woff or woff2. For webfonts, I usually set about 8 kB of kerning weight, which results in 2–3 kB of compressed data size in the woff2.
- Tim Ahrens
- Site Admin
- Posts: 415
- Joined: 11 Jul 2019
Re: Wrong class?
Thanks for the kind words!albertomalossi wrote: ↑28 Dec 2021 Hi Tim,
I understand that it's done with storage space in mind, but this system complicates quite a lot any manual intervention after the fact. Is there a way to achieve "human" classes? Or could it become an option in the future?
BTW the plugin rocks!
a
Sorry, Kern On is not optimized for manually changing the (class) kerning.
The idea is that if you spot anything you are not satisfied with, you simply open Kern On again, add another model, and finalize again. Models (or independent pairs) are the “manual intervention” in the KO system.