# XPixelGroup/HYPIR

Official implementation of HYPIR: Harnessing Diffusion-Yielded Score Priors for Image Restoration (SIGGRAPH 2025)

Repository: https://github.com/XPixelGroup/HYPIR
Canonical: https://ross.abutalabs.com/products/hypir
Language: Python
License: NOASSERTION
License Family: other
Topics: image-restoration, image-super-resolution, image-upscaling
Last push: 2025-10-16T12:15:19+00:00

## Health v2 (maintenance only)
Score: 39/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 47, release rhythm 35, longevity 29
- inputs: {"age_days": 410, "days_push": 321, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1241, forks 102 (observed 2026-08-28T04:04:06.257302+00:00)

## What it is
Official PyTorch implementation of HYPIR, a SIGGRAPH 2025 method that harnesses diffusion-yielded score priors for image restoration. It provides pretrained models (based on Stable Diffusion 2.1) for restoration, super-resolution, and upscaling, with inference and training code.

## Use cases
- restore old or degraded photos with diffusion priors
- upscale and enhance low-resolution images
- super-resolve images to high fidelity
- denoise and repair corrupted images
- train a custom image restoration model
- run image enhancement inference on GPU

## When to choose
- you need state-of-the-art diffusion-based image restoration or super-resolution
- you want pretrained restoration models with inference and training code
- you are doing research on diffusion priors for image enhancement

## When to avoid
- you need fast CPU-only image enhancement without a GPU
- you need a simple classical upscaler like bicubic or Real-ESRGAN with minimal compute
- you cannot accept a non-standard (NOASSERTION) license for commercial use without review

## Facets
- artifact type: library
- maturity: active
- function: image-processing, machine-learning, deep-learning
- domain: image-processing, computer-vision, artificial-intelligence
- platform: python, cross-platform
- tags: diffusion-models, image-restoration, super-resolution, image-upscaling, siggraph-2025, pretrained-models, gpu

## Member repositories
- XPixelGroup/HYPIR (main) score 39

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:06.257302+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T05:08:27.650087+00:00, confidence not recorded.
  - readme: https://github.com/XPixelGroup/HYPIR (fetched 2026-08-28T04:04:06.257302+00:00, sha afb45b7414f5)
- Data as of 2026-08-30T08:39:29.467469+00:00.