Hi author here, I was posting on HN this week to get more interests on my recent progress on reviving zi2zi project.
Motivation behind the project is that I feel font generation has made a long way after zi2zi, but feels still not quite live up to my expectation where it can become a practical technology in creating fonts people can use.
zi2zi-JiT is thus created during last Christmas, aiming to create production grade CJK fonts that can be seen/used in everyday life, instead of simply being a research project.
So far, I have created 2 full set Chinese fonts using zi2zi-JiT, each with 6,763 Chinese characters created (GB2312 standard), from ancient Chinese books/caligraphies:
The project name is obvious to anyone who speaks Chinese, but for everyone else: “zi2zi” means ”字 to 字” or “character to character”, which describes the use of style transfer.
It also makes total sense that the model produces “worse” results when reproducing seal script (an ancient script), because seal script has a lot of variant components that don’t match modern regular script.
Perhaps it would make sense to use an existing seal script font as the content reference when generating seal script output? Maybe this was done and the model is just not tuned for it?
Motivation behind the project is that I feel font generation has made a long way after zi2zi, but feels still not quite live up to my expectation where it can become a practical technology in creating fonts people can use.
zi2zi-JiT is thus created during last Christmas, aiming to create production grade CJK fonts that can be seen/used in everyday life, instead of simply being a research project.
So far, I have created 2 full set Chinese fonts using zi2zi-JiT, each with 6,763 Chinese characters created (GB2312 standard), from ancient Chinese books/caligraphies:
https://github.com/kaonashi-tyc/Zi-QuanHengDuLiang
https://github.com/kaonashi-tyc/Zi-XuanZongTi
All created in matters of 2-3 days, and free for commercial uses.
There are still some issue with current approach, but I am happy to hear you guys feedbacks and improve from here :)
The project name is obvious to anyone who speaks Chinese, but for everyone else: “zi2zi” means ”字 to 字” or “character to character”, which describes the use of style transfer.
It also makes total sense that the model produces “worse” results when reproducing seal script (an ancient script), because seal script has a lot of variant components that don’t match modern regular script.
(The seal script image: https://github.com/kaonashi-tyc/zi2zi-JiT/blob/main/assets/g... )
Perhaps it would make sense to use an existing seal script font as the content reference when generating seal script output? Maybe this was done and the model is just not tuned for it?
The content encoder, the structure prior in this model is based on SongTi(SourceHans Serif), seal script would be foreign to it
http://ns2.tug.org/TUGboat/tb24-1/yiu.pdf
i think this is a dos port of it? http://www.cbflabs.com/down/show.php?id=62
seems to work for me in dosbox