Skip to content

Thera

Arbitrary-scale super-resolution without aliasing

Visit official site

What Thera is

The tool is described as a super-resolution technique that produces high-quality images at different scales, with a built-in physical observation model that avoids aliasing. It was developed by a research team at ETH Zurich for image enhancement and computer vision, especially remote sensing and photogrammetry.

What you can do with it

  • Generate images at arbitrary super-resolution scales
  • Use a built-in physical observation model to avoid aliasing
  • Choose among several pre-trained models
  • Run model evaluation and training
  • Work from a simple command-line interface

Application scenarios

  • Apply it in remote sensing to improve ground object recognition
  • Raise photo quality in digital image processing software
  • Provide high-resolution images for photogrammetry and geographic analysis

Getting started

  • Ensure Python 3.10 and an NVIDIA GPU are installed
  • Download the required pre-trained model files
  • Install the needed dependencies with pip
  • Run the super-resolution script from the command line with input and output files

Pros & cons

✓ What we like

  • Runs from a simple command-line interface
  • Supports model evaluation and training

! What to watch out for

  • Python 3.10 and an NVIDIA GPU are required
  • Pre-trained model files must be downloaded before running

FAQ

What keeps the output free of aliasing?

The material points to a built-in physical observation model that avoids aliasing.

How is the tool run?

Through a simple command-line interface with input and output files.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review