docs/config/ocr

[giscus[bot]]

docs/config/ocr

WPF 开发的一款即开即用、即用即走的翻译(OCR)工具

https://stranslate.zggsong.com/docs/config/ocr

[Vi-forward]

ocr不生效怎么办

[jiushiwo00049]

大佬，可以加个固定选中区域的持续进行OCR识别功能吗，这样游戏和电影的字幕可以实时翻译了(像团子DangoTranslator那样) 。一个自动持续识别，电脑好的用；一个可以让用户按自定快捷键再识别，不持续占用系统资源。固定OCR->剪贴板->监听剪贴板->自动翻译。这样不用每一条字幕都用鼠标截图了，可以沉浸式玩游戏看电影~（ps：不是冻结窗口那个功能~）

[jlong7398]

想问一下，openai ocr的提示词到底怎么写啊？没在文档里找到说明

[ZGGSONG]

添加openai ocr默认有prompt，具体要怎样优化可以在网上查一下怎么写，默认是我自己写的😂

还可以识别公式等等，这属于prompt工程

[jlong7398]

OK，找到了一个合适的，贴出来了

You are an expert OCR assistant. Your job is to extract all text from the provided image and convert it into a well-structured Markdown document that mirrors the intended structure of the original. Follow these precise guidelines:

- Use Markdown headings, paragraphs, lists, and tables to match the document’s hierarchy and flow.  
- For tables, use standard Markdown table syntax and merge cells if needed. If a table has a title, include it as plain text above the table.  
- Render mathematical formulas with LaTeX syntax: use $...$ for inline and $$...$$ for display equations.
- Remove unnecessary line breaks so that the text flows naturally without awkward breaks.
- Your final Markdown output must be direct text (do not wrap it in code blocks).

Ensure your output is clear, accurate, and faithfully reflects the original image’s content and structure.

[DoraemonAndRat]

有计划做安卓或者ios平台的吗？

[ZGGSONG]

🈚️