- Home
- AI Video Restoration Software
- RestoraX
RestoraX
What it is
RestoraX is a free, open-source toolkit for restoring old films, home videos, and archival footage with video and audio processing models. The project describes 25 models for tasks including super-resolution, colorization, face restoration, frame interpolation, scratch and dust removal, deinterlacing, stabilization, SDR-to-HDR conversion, and audio restoration. Its visual pipeline builder connects processing stages through branches, merge strategies, retries, and progress tracking. RestoraX provides a React web interface, a FastAPI REST API with WebSocket progress, a command-line interface, and a ComfyUI node pack. It is designed for self-hosting, with Docker Compose setups documented for development and production. Plugins distributed as separate PyPI packages can add third-party restorers, and model weights download from HuggingFace Hub on first use. The project uses the MIT License. Minimum requirements include Python 3.11, 8 GB RAM, 5 GB disk space, and FFmpeg. CPU processing works; the project recommends CUDA 12.1 or later with at least 8 GB of GPU VRAM. The maker names anime, VHS, and newsreel restoration among its use cases.
For and against
For
- Free open-source software under the MIT License
- Describes 25 restoration models
- Provides web, REST API, CLI, and ComfyUI interfaces
- CPU processing is supported
Against
- Requires Python 3.11 and FFmpeg
- Minimum stated memory is 8 GB RAM
- CUDA 12.1 or later with 8 GB GPU VRAM is recommended
The verdict
RestoraX offers a broad set of restoration models and several ways to build or run processing workflows. It is self-hosted, so users should account for its stated system requirements and deployment setup.
Who it suits
RestoraX may suit people restoring archival footage, home videos, old films, anime, VHS, or newsreels who can run self-hosted software. It offers web, API, and command-line interfaces for different workflows.
Plans and prices
RestoraX
Open-source MIT software · self-hosted · GPU/CPU processing
Read 3 Oct 2026 · github.com
On the record
- ComfyUI
- The project describes a ComfyUI node pack as part of its platform.github.com · read 3 Oct 2026
- Deployment
- The maker documents Docker Compose setups for development and production and describes the software as self-hostable.github.com · read 3 Oct 2026
- Extensibility
- Third-party restorers can be added through plugins distributed as separate PyPI packages.github.com · read 3 Oct 2026
- Integrations
- The documented stack uses Celery and Redis for background jobs, PostgreSQL or SQLite for data storage, and MinIO in the production Docker setup.github.com · read 3 Oct 2026
- Intended users
- The README names old-film, home-video, archival-footage, anime, VHS, and newsreel restoration as use cases.github.com · read 3 Oct 2026
- Interfaces
- RestoraX provides a React web UI, a FastAPI REST API with WebSocket progress, and a command-line interface.github.com · read 3 Oct 2026
- License
- The repository is released under the MIT License, which grants use, modification, and distribution subject to its terms.github.com · read 3 Oct 2026
- Model weights
- Model weights download automatically from HuggingFace Hub on first use.github.com · read 3 Oct 2026
- Pipeline builder
- Its visual pipeline builder uses a DAG engine with typed ports, parallel branches, merge strategies, retry policies, and per-branch progress.github.com · read 3 Oct 2026
- Purpose
- RestoraX is an open-source AI video and audio restoration toolkit for old films, home videos, and archival footage.github.com · read 3 Oct 2026
- Requirements
- The stated minimums include Python 3.11, 8 GB RAM, 5 GB disk, and FFmpeg; the page says CPU works and recommends CUDA 12.1 or later with 8 GB or more of GPU VRAM.github.com · read 3 Oct 2026
- Restoration models
- The project describes 25 models for super-resolution, colorization, face restoration, frame interpolation, scratch and dust removal, deinterlacing, stabilization, SDR-to-HDR conversion, and audio restoration.github.com · read 3 Oct 2026
- Security and compliance
- The opened maker pages provide an MIT license but do not state a security certification or compliance standard.github.com · read 3 Oct 2026
