Greasy Fork is available in English.
(WIP) Translate texts in images on Pixiv, Twitter, Misskey, Calckey
The match all url description in the script is just because some sites don't have a consistent url, it doesn't mean it'll correctly run on all sites. If you're uncomfortable with that, just remove the match all url line in your userscript manager, and the script will still work on other listed sites.
Userscript for https://cotrans.touhou.ai/. (WIP)
Translate texts in images with support for:
If the website you want isn't listed above and you'd like to translate an image from the site, consider directly uploading to https://cotrans.touhou.ai/ to get your image translated.
If you find this script helpful, please consider supporting us on ko-fi / Patreon!
Translation samples: https://github.com/zyddnys/manga-image-translator/blob/main/README.md#samples
Project source code: https://github.com/VoileLabs/cotrans
Notes:
This is an automatic uploader, not a standalone translator. Translations are submitted anonymously and performed on the server.
If this script requests to access a URL, please allow it. It's just the script trying to download images from the website's CDN.
This script requires browsers that are compatible with ES2020, or more specifically:
Chrome/Edge 80+, Safari 14.1+, Firefox 80+, Opera 67+
If you can't use one of these browsers, contact me with a reason. Legacy browser support takes a lot of time, and I'd like to not do it.This script requires
HTMLCanvasElement.toBlobsupport for resizing large images above 4k. If you have something that disables this feature, eg. Tor, then large images will result in errors.
The script itself (interface): 简体中文 (Simplified Chinese), English.
Source language (language of the original image): 中文 (Chinese), 日本語 (Japanese), 한국어 (Korean), English.
Target language (language of the translated image): 简体中文, 繁體中文, 日本語, English, 한국어, Tiếng Việt, čeština, Nederlands, français, Deutsch, magyar nyelv, italiano, polski, português, limba română, русский язык, українська мова, español, Türk dili.
Absolutely! Just change the translator to None (remove texts) and you'll get an image with all text removed.
Settings are directly embedded into target websites'(listed above) settings.
You can find the userscript settings under website's settings, or you can use these direct links:
Userscript settings are synced globally across all websites.
You can try increasing "Text detection resolution":
Changing the setting will result in a longer processing time with better accuracy.
If the text is still not detecting, try changing to a different text detector and see if it helps.
There are also some issues with handwritten(-like) texts, those texts are hard to do OCR with, sorry.
Sorry, but no machine translator is perfect.
You can try another translate service which can be changed in script settings.
If you think it's probably to do with text recognition, you can try bumping the direction resolution or changing to another detecter.
You can change "Render text orientation" in settings to "Horizontal only" if that helps.
