traiNNer-redux
Deep learning framework to train PyTorch models for image and video upscaling.
Deep learning training framework for image and video super resolution in PyTorch.
traiNNer-redux is a specialized deep learning training framework designed for creating custom PyTorch models focused on image super resolution and restoration. It is built for graphic designers, digital artists, and technical creators who want to go beyond pre-trained AI models and train their own networks.
What it is
The framework allows you to train PyTorch models for upscaling and restoring images and videos using various architectures imported from Spandrel and developed by contributors like umzi2 and Kim2091. It integrates with tools like OpenModelDB for pretrain models and chaiNNer for dataset preparation and running trained models.
Who it's for
It suits digital artists and technical creators who need custom upscaling models and have the hardware to handle deep learning training. It does not suit casual users looking for a one-click desktop app without technical overhead.
Key features
PyTorch Training Framework
Train custom models for image and video super resolution and restoration.
Spandrel Architectures
Supports a wide variety of advanced network architectures imported via Spandrel.
OpenModelDB Integration
Use existing AI models as pretrain models and submit your own finished models.
Dataset Preparation Tools
Leverage companion utilities like WTP Dataset Destroyer and Video Destroyer to degrade high-quality sources.
Up and running in 3 steps
Our honest verdict
What's great
- Supports custom neural network architectures like ArtCNN, FlexNet, and TSPANv2
- Compatible with chaiNNer for pipeline integration and dataset prep
- Works with pre-trained models from OpenModelDB to jumpstart training
- Includes support for video-focused architectures with temporal consistency
Where it falls short
- Requires a capable NVIDIA graphics card, or an AMD card running Linux with ROCm
- Involves deep learning training workflows rather than simple point-and-click upscaling
- Setup requires following documentation and managing Python environments
Pricing & license
Free under its open source license. The only cost is running it yourself, which includes the electricity and hardware required to train deep learning models.
Open remote AI-training roles
767 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
What hardware do I need to run traiNNer-redux?
An NVIDIA graphics card is strongly recommended, though AMD graphics cards are supported on Linux machines using ROCm.
Can I use models trained here in other software?
Yes, models trained with this repository can be run in chaiNNer and submitted to OpenModelDB.
Where do I get pretrain models?
You can source pretrain models from OpenModelDB to help train your new models.
