gitwtfhub

wtf is real-sr?

tencent/real-sr — explained in plain English

Analysis updated 2026-07-17 · repo last pushed 2022-01-16

★ 824PythonAudience · researcherComplexity · 4/5DormantSetup · moderate

TL;DR

RealSR sharpens blurry, noisy real-world photos by learning how actual camera degradation happens, unlike tools trained only on artificially blurred images, it won CVPR NTIRE 2020's top prize.

Mindmap

mindmap
  root((repo))
    What it does
      Sharpens real photos
      Removes noise and blur
      Learns real degradation
    Tech stack
      Python
      PyTorch
      GPU
    Use cases
      Enhance phone photos
      Fix compressed JPEGs
      Restore old images
    Audience
      Researchers
      Photo enthusiasts

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Why would anyone build with this?

REASON 1

Sharpen blurry or noisy phone photos using a pre-trained model without any coding.

REASON 2

Restore compressed JPEG images with visible processing artifacts.

REASON 3

Train a custom super-resolution model on your own dataset of real-world degraded images.

REASON 4

Batch-process a folder of low-quality images into enhanced versions.

What's in the stack?

PythonPyTorch

How it stacks up

tencent/real-srgvclab/cutclawjd-opensource/joyai-echo
Stars824831758
LanguagePythonPythonPython
Last pushed2022-01-16——
MaintenanceDormant——
Setup difficultymoderatemoderatehard
Complexity4/53/55/5
Audienceresearchergeneralresearcher

Figures from each repo's GitHub metadata at analysis time.

How do you spin it up?

Difficulty · moderate Time to first run · 30min

GPU recommended for efficient inference, standalone executables are available for non-coders.

Yoink these prompts

Prompt 1
Show me how to run RealSR's pre-trained model on a folder of my own low-quality photos.
Prompt 2
Explain how RealSR estimates real blur kernels instead of using synthetic degradation like other super-resolution tools.
Prompt 3
Walk me through preparing a dataset and training my own RealSR model from scratch.
Prompt 4
How do I use RealSR's standalone executable to enhance photos without setting up Python or PyTorch?

Frequently asked questions

wtf is real-sr?

RealSR sharpens blurry, noisy real-world photos by learning how actual camera degradation happens, unlike tools trained only on artificially blurred images, it won CVPR NTIRE 2020's top prize.

What language is real-sr written in?

Mainly Python. The stack also includes Python, PyTorch.

Is real-sr actively maintained?

Dormant — no commits in 2+ years (last push 2022-01-16).

How hard is real-sr to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is real-sr for?

Mainly researcher.

View the repo → Decode another repo

This repo across BitVibe Labs

Don't trust strangers blindly. Verify against the repo.

Other repos in this lane