[novok]

A halfway tool that would let you take the text and auto-typeset it in would probably be a big thing. All you would need is the equivalent of a subtitles file at that point. The tool will identify regions to change and how to change them, the original text and the machine translation text. Then, all you would need is somebody who can do translations properly afterward to give a better translation, and with less friction, it will happen a lot more. And you could have a user rating of which translation is better.

I could also imagine a tool that would take a page and describe per panel. Then, the translation files with LLMs could have a better idea of the context and probably make a better translation versus a speech bubble by speech bubble type translation since they would only interact with the intermediate 'subtitles' file. It would also be a good accessibility thing for blind people to read a comic.