Skip to content
VSR
Removing hard-coded subtitles from videos and text watermarks from images with lossless resolution
Features
System Requirements
Minimum 8GB RAM. 9GB+ storage recommended.
macOS 15+: M-series chips required.
Windows 10/11: CPU supported, but NVIDIA GPU recommended.
Note: For NVIDIA GPUs, install a newer driver.Introduction
Video Subtitle Remover(VSR) is an AI-based local tool designed to remove hard-coded subtitles from videos and text-like watermarks from images, generating output files with lossless resolution. It requires no third-party API, enabling all functions to be implemented locally for data privacy and efficiency.
Core Functions
- Hard Subtitle/Text Watermark Removal：Utilizes AI algorithms to accurately identify and remove hard-coded subtitles from videos and text watermarks from images, producing clean files.
- Intelligent Area Inpainting：Employs advanced AI models to fill in the removed subtitle/watermark areas with pixel-level precision, avoiding mosaic or blurry effects.
- Flexible Customization Modes：
- Manual position specification for precise subtitle removal;
- Automatic full-video scanning for batch text element processing.
- Batch Processing：Supports batch watermark removal for multiple images, enhancing workflow efficiency.
Technical Features and Advantages
- Multi-platform Compatibility：
- Pre-built Windows packages (supporting NVIDIA CUDA and AMD/Intel DirectML acceleration);
- Docker deployment available for various GPU models (e.g., Nvidia 10/20/30/40/50 series, AMD/Intel discrete/integrated graphics).
- High-performance Algorithms：
- Integrates STTN, LAMA, PROPAINTER, and other AI algorithms optimized for different scenarios:
- STTN: Ideal for live-action videos, fast speed with optional detection skip;
- LAMA: Superior for images and animated videos;
- PROPAINTER: Suited for high-motion videos with excellent results.
- Integrates STTN, LAMA, PROPAINTER, and other AI algorithms optimized for different scenarios:
- Local Operation：Eliminates reliance on third-party APIs, ensuring data security and stable performance without network dependency.
- Open-source and Scalable：Provides source code and training tools, allowing users to customize models for improved removal effects.
Application Scenarios
- Video post-production: Removing hard subtitles from film clips or educational videos;
- Image watermark cleaning: Batch processing watermarked material images;
- Content re-creation: Providing subtitle-free raw files for translation, re-dubbing, etc.
LicenseApache-2.0