# zsyOAOA/InvSR

Arbitrary-steps Image Super-resolution via Diffusion Inversion (CVPR 2025)

Repository: https://github.com/zsyOAOA/InvSR
Canonical: https://ross.abutalabs.com/products/invsr
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-02-07T23:01:09+00:00

## Health v2 (maintenance only)
Score: 51/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 66, release rhythm 35, longevity 45
- inputs: {"age_days": 634, "days_push": 207, "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 1443, forks 93 (observed 2026-08-28T04:04:44.885624+00:00)

## What it is
InvSR is a Python research library implementing arbitrary-steps image super-resolution via diffusion inversion, leveraging pre-trained diffusion models with a partial noise prediction strategy. It supports flexible sampling from one to five steps and ships with Gradio, Hugging Face, Replicate, and Colab demos.

## Use cases
- upscale low-resolution images with diffusion models
- one-step image super-resolution
- restore blurry or degraded photos to high resolution
- run super-resolution with a configurable number of sampling steps
- experiment with diffusion inversion for image restoration
- deploy an image upscaling demo on Hugging Face or Replicate

## When to choose
- you need state-of-the-art diffusion-based super-resolution with fast, few-step sampling
- you want a research-grade SR model with ready-made online demos
- you have a GPU and want flexible control over the quality/speed trade-off

## When to avoid
- you need a lightweight CPU-only upscaler
- you require a permissively licensed production library (license is non-standard)
- you only need simple classical upscaling like Lanczos or ESRGAN

## Facets
- artifact type: library
- maturity: active
- function: image-processing, machine-learning, deep-learning, llm-inference
- domain: image-processing, computer-vision, deep-learning, artificial-intelligence
- platform: python, cross-platform
- tags: super-resolution, diffusion-models, diffusion-inversion, cvpr-2025, image-restoration, research-code, gradio-demo, gpu, linux

## Member repositories
- zsyOAOA/InvSR (main) score 51

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:44.885624+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 2026-08-30T04:36:22.665910+00:00, confidence not recorded.
  - readme: https://github.com/zsyOAOA/InvSR (fetched 2026-08-28T04:04:44.885624+00:00, sha 7bacc51e2e01)
- Data as of 2026-08-30T08:39:29.467469+00:00.