Overview

This application captures text from your screen, recognizes it, and translates it.

Generative AI is used for text recognition and translation.

It is designed primarily for translating text from games with low motion.

Screen capturing and translation are performed in a window separate from the target window.

Table of Contents

• System Requirements
• Prerequisites
• Initial Setup
  • Generative AI Setup
  • Translation Setup
  • Block Detection Setup
  • UI Language Setup
• Usage
  • Select Target Window
  • Region Translation
  • Save Region and Translate
  • Full Screen Translation
  • Block Translation
  • Translate Image from Clipboard
  • Re-run Translation Only
  • Cancel Operation
• Reference: Verified Generative AIs
  • LM Studio
  • Gemini API

System Requirements

• Windows 11

Prerequisites

• Generative AI environment (Generative AI is not included with this application)
  • LM Studio was used for testing during development.

Initial Setup

Settings to perform once initially.

Generative AI Setup

1. Open the myTranslator settings screen with  and select AI.
2. Set the OCR endpoint.
   OpenAI-compatible APIs can be used.
3. Click the "Get Models" button and select a model.
   If using a paid model, set your API Key first.
4. Configure translation settings in the same manner.
   If using the same settings as OCR, check "Same as OCR".
5. If using the Block Translation feature, configure Block Detection in the same manner.

Translation Setup

Configure these settings when customizing translation behavior, such as changing the output language.

1. Open the myTranslator settings screen with  and select Translation.
2. Enter the prompt instructing translation in "Translation Instructions".
   

Block Detection Setup

Configure settings for detecting blocks when using the Block Translation feature.

1. Open the myTranslator settings screen with  and select Block Detection.
2. Specify the coordinate range used for Bounding Boxes in "Image Size".
   This varies depending on the model used. Check the model's documentation for specific values.
3. Specify the coordinate type used for Bounding Boxes in "Coordinate Type".
   This varies depending on the model used. Check the model's documentation for specific values.

UI Language Setup

Configure the language of the myTranslator UI.

1. Open the myTranslator settings screen with  and select UI.
2. Select the language from the "Language" dropdown.

Usage

Select Target Window

• Open the window selection screen with  and select the target window.
  If the target application is not running, launch the application first, then click the "Refresh" button.

Region Translation

Translates text within a specified region.

1. Open the capture viewer screen with  and select the region you want to translate by dragging with the mouse.
2. Click the "Region Translation" button to perform text recognition and translation on the selected region.
   If "Auto close after selection" is checked, text recognition and translation will be performed automatically after selecting the region.

To translate the same region again, press .

Save Region and Translate

Saves the region selected in Region Translation.

1. Open the translation region screen with .
2. Specify the save location and name, then click the "OK" button.

Select a saved region from the toolbar dropdown list and press  to translate.

To delete or rename settings, right-click on the target item in the dropdown list and use the menu.

Full Screen Translation

Translates text across the entire screen.

• Translate the entire screen with .

Block Translation

Translates screen text block by block. Displays translation results on top of the image in the capture viewer screen.

• Perform block translation with .

Translate Image from Clipboard

1. Open the capture viewer screen with .
2. Click the "Paste" button to paste an image from the clipboard.
3. Click one of the buttons: "Region Translation", "Full Translation", or "Block Translation".

Re-run Translation Only

If part of the text recognition fails or is inaccurate, manually edit the recognized text and re-run translation.

1. Edit the recognized text.
2. Translate with .

Cancel Operation

• Cancel operation with .

Reference: Verified Generative AIs

LM Studio

Endpoint: http://localhost:1234/v1

Gemini API

Endpoint: https://generativelanguage.googleapis.com/v1beta/openai