ocr flat-out doesn't work.

[Zoobastix]

I've imported first volume of Ajin with a zip file. However, the ocr doesn't work, it didn't recognize any text, nothing. It is enabled. Am I supposed to do something in addition to just importing manga to a website?

[Gnathonic]

You have to import the .mokuro file with the zip. If you don't have a .mokuro file you have to make one.
Mokuro-reader doesn't OCR volumes. It can show the OCR created by Mokuro, or you can use Mokuro-bunko (still in alpha).
Mokuro is a python script you run on your computer before adding your manga to mokuro-reader.
Mokuro-bunko is a WebDAV server that will add OCR to volumes after you've added them to mokuro-reader and backed them up to your mokuro-bunko instance.

[Zoobastix]

I see, I didn't realize that. I didn't see this requirement in the quick start section.

Can you help me with something else? I installed python and mokuro but now it's stuck on
pytorch_model.bin: reconstructing file: 0%| | 0.00B / 444MB
when I run it on a manga volume.

[Gnathonic]

No, I can't help you. Ask an LLM to help you get it set up. They'll read the docs and the logs if needed. They are pretty good at fixing python related problems.
Plus, Mokuro isn't my project.
If something is actually wrong with it, I'm the wrong person to help you with that.

I maintain mokuro-reader (this repo), and mokuro-bunko.

[Zoobastix]

Which LLM?

[Gnathonic]

Claude Code, ChatGPT Codex, or Grok Build should all work. I use Claude Code. A lot of folks get good results with Codex. Grok Build is the new one, and probably can do this just fine.
Use whichever one you are already paying for. You can't use any of these for free.

[Zoobastix]

I don't even want to use LLMs, let alone pay for them.

[Gnathonic]

Your other option is to read the instructions. Pick your poison.

[Zoobastix]

The only instruction you've given is:

Process your manga with mokuro (see requirements below)
Upload the processed folder or ZIP containing your manga images and the .mokuro file
Start reading!

and as for Mokuro itself it doesn't mention my issue at all. It just assumes that there will be no problems like that.
I did it according to instruction, it just fails to proceed at the manga processing step.

[Zoobastix]

I'm not a techie, I'm just looking for help with this language learning aid. No need to treat me like an idiot.

[Gnathonic]

Mokuro has an issues area, with reported issues. It also has a detailed guide, which it does mention on the github page.

If you don't want to read just get an LLM subscription and tell it to read and execute for you. Those things are thousands of times faster than we are, and with how bad search engines are nowadays, they are quite useful.

But to each their own.

You could try mokuro-bunko. The windows install just got an upgrade, and it should work out of the box if you follow the instructions. Most of the "alpha" stuff are features you don't need if you just want to OCR some stuff on the same computer. It is still techie oriented if you want everything it offers, but it should be easy enough to use for the basics.

Or you could try Chimahon or Manatan. They have built in OCR, but aren't web apps.