A couple of questions

[bropines]

The program when performs all translations, painting over even what is not needed. Like I don't want any cloud to be painted over and translated, with text, but it can't be undone. There is no paint layer as such. I would like to be able to remove probably painted and translated pieces

If you don't understand anything, I can attach a video)

[dmMaze]

Did you mean manually perform/correct text removal(i.e., inpainting)? Actually there is a inpainted layer, it just doesn't show explicitly in the ui. And there is a mask layer to show which part are inpainted. To 'erase' unwanted inpainted results, use the inpainted tool or rect tool with your right button pressed, please refer to the following gifs and notice the keystrokes are shown on the top left corner:

https://github.com/dmMaze/BallonsTranslator/blob/master/doc/src/imgedit_inpaint.gif https://github.com/dmMaze/BallonsTranslator/blob/master/doc/src/rect_tool.gif

[bropines]

It worked, thanks Does the project have its own discord/telegram to throw off bugs and error reports? Github allows it, but often bugs are simply associated with the departure of python modules.

[dmMaze]

Does the project have its own discord/telegram to throw off bugs and error reports?

Nope.

Github allows it, but often bugs are simply associated with the departure of python modules

I'll add some instructions to readme for running the source code.

[bropines]

One more question before closing the issue. Can you add a config to translate to another language? I would like to translate the program settings into my language, but I don't really want to get into the code.

[dmMaze]

This program can be translated using linguist.exe, which should be installed with pyqt.

If you are comfortable with it, edit ballontranslator/data/translate/template.ts in linguist and export to ballontranslator/data/translate/[LANGUAGE].qm

You can print QLocale.system().name() in the main function of ballontranslator/__main__.py to determine the [LANGUAGE] of your system:

Please notice the template.ts is a bit out of date, I'll update it after I merge the textlayout branch, perhaps later this week.

[dmMaze]

edit ballontranslator/data/translate/template.ts

Update: run python ballontranslator\scripts\update_translation.py to generate or update [LANGUAGE].ts

[bropines]

edit ballontranslator/data/translate/template.ts

Update: run python ballontranslator\scripts\update_translation.py to generate or update [LANGUAGE].ts

Thanks

[bropines]

A couple more questions.

1. Is it possible to add support for yandex translate?
2. Is it possible to translate individual clouds as in this project (https://github.com/AntonRls/Manga-Translator)?
3. In new versions of the code (if you just run it with all python modules), the GPU does not work. Just trying to poke at the video card (and I have as many as 2 of them from AMD and Nvidia), nothing happens

[dmMaze]

A couple more questions.

1. I don't have an access key required by yandex, however, if you're familar with python and know how to use yandex's api, it would be easy to add it to the program, you can refer to https://github.com/dmMaze/BallonsTranslator/blob/master/doc/%E5%8A%A0%E5%88%AB%E7%9A%84%E7%BF%BB%E8%AF%91%E5%99%A8.md and this pr
2. You mean ocr&translate single textblock created by user? No, I'll consider adding it to the right-click menu in the future. By the way, all translated texts are interactly editable.
3. try pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116, AMD GPU would not work on windows.

[bropines]

When you click once on a spot with the text-coloring tool, it gives an error. Just right-click on the spot

[bropines]

And let's say. I accidentally deleted a block.

1. I did not understand how to make mine with text.
2. Is it possible to apply OCR and all operations to only one page?

[bropines]

And I still don't have cuda support. I have two video cards. One is built-in, the second is discrete. Maybe because of this, cuda does not work?

[Snowad14]

Use your nvidia card, go to your terminal, type "python" then "import torch" and "torch.cuda.is_available()" and make sure it returns True otherwise your installation of pytorch is not well installed or you do not have install cuda on your system " nvcc --version" to see if you have installed cuda well. Also, why not use the compiled version you won't have any problem I think

[dmMaze]

When you click once on a spot with the text-coloring tool, it gives an error. Just right-click on the spot

fixed d55a1a054b2734dac46e9af8c9bdb138a467143f

1. I did not understand how to make mine with text

Activate text editing mode and the rectangle button next to it:

then drag mouse on the canvas with right button pressed, see following gif, it's after delete operation

2. Is it possible to apply OCR and all operations to only one page?

No

[bropines]

Now I think my friend and I will make a couple of variants of the logo (icon). I'll send the options here.