zsyOAOA/ResShift
ResShift is an efficient diffusion model for image super-resolution that reduces inference steps by shifting residuals between high-resolution and low-resolution images.
Not currently ranked — collecting fresh signals.
star history
This repository implements a novel diffusion-based super-resolution method that constructs a Markov chain to transfer between high-resolution and low-resolution images through residual shifting, substantially improving transition efficiency. It includes a custom noise schedule for controlling shifting speed and noise strength during the diffusion process. The work achieved NeurIPS 2023 Spotlight and TPAMI 2024 publication, with code, pre-trained models, and Google Colab support.
Frequently asked
- What is zsyOAOA/ResShift?
- ResShift is an efficient diffusion model for image super-resolution that reduces inference steps by shifting residuals between high-resolution and low-resolution images.
- Is ResShift open source?
- Yes — zsyOAOA/ResShift is an open-source project tracked on heatdrop.
- What language is ResShift written in?
- zsyOAOA/ResShift is primarily written in Python.
- How popular is ResShift?
- zsyOAOA/ResShift has 1.4k stars on GitHub.
- Where can I find ResShift?
- zsyOAOA/ResShift is on GitHub at https://github.com/zsyOAOA/ResShift.