MashiroSaber03/Saber-Translator

Features

• Manga Translation Pipelines - Provides a specialized workflow for detecting text bubbles and translating manga imagery while preserving layout and reading direction.
• AI-Powered Translations - Implements AI-driven linguistic translation of text extracted from images using large language models.
• Background Cleaning - Removes original text from images using intelligent inpainting or solid color fills to prepare for new text rendering.
• Manga Analysis Engines - Extracts plot points, character guides, and story summaries using vector search and semantic indexing.
• Long Multimodal Contexts - Processes groups of images using long multimodal contexts to maintain character and story consistency.
• Multimodal Translation - Uses multimodal AI to analyze multiple images simultaneously, ensuring narrative and character coherence across manga pages.
• Translation Editors - Provides a rich-text interface specifically for refining and styling translated content before final rendering.
• Narrative Consistency Analysis - Uses AI to analyze characters and plot across multiple pages to maintain narrative consistency and generate summaries.
• Visual Typography Editors - Provides a visual interface for refining translated text, fonts, and bubble positions to ensure professional typography.
• Visual Editor Interfaces - Provides a graphical editing area for manually refining OCR bounding boxes, font styles, and text placement.
• AI Story Summarizers - Generates structured overviews of plot, characters, and world-building using AI-driven synthesis.
• Manual Bounding Box Refinement - Allows manual adjustment of font, size, color, and position to correct automated text detection.
• Image Inpainting - Utilizes generative neural networks to fill in erased text areas and restore original manga backgrounds.
• Batch Processing - Provides automated execution of translation pipelines across multiple image files to maintain narrative consistency.
• Narrative Intelligence Engines - Extracts characters, plot points, and dialogues using a hierarchical engine that summarizes content from page to book level.
• Multi-Type Step Chains - Implements a chain of modular translation tasks where each step passes its output as input to the next.
• Story Timeline Mapping - Generates structured event sequences and character action lines based on the analyzed progression of the story.
• Real-Time Layout Synchronization - Synchronizes text styles and bubble placements between the image view and a sidebar editor for real-time rendering.
• Narrative Content Querying - Answers questions about specific plot points or general themes using vector search and direct page references.
• Vector-Database-Backed Retrievals - Uses vector database indices to perform semantic similarity searches for querying story content and events.
• Narrative Content Aggregators - Summarizes manga data from a granular page level up to a global book level to generate plot timelines.
• Visual Semantic Search - Locates specific scenes or topics using vision-language models to index and retrieve visual content.
• Semantic Search - Finds specific scenes or plot points across a library using vector search and citations.
• Extensible Plugin Architectures - Provides an extensible architecture that allows users to inject custom logic into specific pipeline stages.
• Plugin-Based Logic Extensions - Provides a plugin-based architecture for adding custom logic and configuration to manga processing workflows.
• Bidirectional Canvas Synchronizations - Links a visual image canvas with a sidebar editor to instantly reflect text and layout changes in the preview.
• Translation Workflow Orchestrators - Implements a plugin-based system for defining complex, multi-step translation sequences and orchestration.

Star history