Skip to content

Add Gemini OCR plugin#49

Open
Bl0ck154 wants to merge 1 commit into
pot-app:mainfrom
Bl0ck154:add-gemini-ocr-plugin
Open

Add Gemini OCR plugin#49
Bl0ck154 wants to merge 1 commit into
pot-app:mainfrom
Bl0ck154:add-gemini-ocr-plugin

Conversation

@Bl0ck154

@Bl0ck154 Bl0ck154 commented May 16, 2026

Copy link
Copy Markdown

Summary

  • Add pot-gemini-ocr-patched to the recognize plugin list.
  • Include entries in list.json, README.md, and README_EN.md.

Why another Gemini OCR plugin?

There is already a community Gemini OCR plugin, but this patched variant is aimed at a cleaner out-of-the-box experience for non-Chinese users and current Gemini Developer API usage.

Compared with the existing community plugin, this version changes:

  • English UI labels/descriptions for plugin settings instead of Chinese-only labels such as 流式输出, 模型选择, 自定义模型名称, 推理长度, and 其他参数.
  • Updated practical model presets based on ModelService.ListModels, including current Gemini 3.1/2.5/2.0 options and latest aliases.
  • Removed older preview/experimental presets such as gemini-2.5-flash-preview-04-17 and gemini-2.5-pro-exp-03-25 from the main preset list.
  • Uses Custom model name as the recommended path so users can paste future Gemini model IDs without waiting for plugin updates.
  • Accepts both gemini-* and models/gemini-* custom model formats by normalizing the models/ prefix.
  • Streaming is disabled by default because normal generateContent response parsing is simpler and more reliable for OCR in Pot.
  • Adds an OCR-specific prompt that preserves original language, alphabet, punctuation, casing, and line breaks, and prevents translation/explanations.
  • Adds Gemini 3.1 thinkingLevel: MINIMAL guidance via generationConfig for faster OCR-oriented responses.
  • Provides English install/build documentation and a ready-to-install .potext file in GitHub Releases.

Tested

The plugin was installed and tested in Pot on Windows with Gemini Developer API. Screenshot OCR works correctly with multilingual text using this setup:

  • API Base URL: https://generativelanguage.googleapis.com/v1beta
  • Model: gemini-3.1-flash-lite
  • Streaming: No
  • Extra generationConfig JSON: {thinkingConfig:{thinkingLevel:MINIMAL}}

Plugin

https://github.com/Bl0ck154/pot-gemini-ocr-patched

This is meant as a cleaned-up, tested Gemini OCR option for users who need reliable multilingual screenshot OCR in Pot.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant