kornia/real_esrgan
Pretrained weights for Real-ESRGAN, used by kornia's RRDB super-resolution models
(kornia.contrib.super_resolution.RRDBNetBuilder).
Real-ESRGAN is a blind super-resolution network, an RRDB generator trained on purely synthetic degradations. ICCV Workshops 2021.
Original repo: xinntao/Real-ESRGAN
Weights
| File | Model | Scale | RRDB blocks |
|---|---|---|---|
RealESRGAN_x4plus.safetensors |
RealESRGAN_x4plus | x4 | 23 |
RealESRNet_x4plus.safetensors |
RealESRNet_x4plus | x4 | 23 |
RealESRGAN_x4plus_anime_6B.safetensors |
RealESRGAN_x4plus_anime_6B | x4 | 6 |
RealESRGAN_x2plus.safetensors |
RealESRGAN_x2plus | x2 | 23 |
Provenance
Each file holds the params_ema state dict of the upstream release asset with the same stem, unchanged: the same
keys, dtypes, shapes and bits. That state dict is what the upstream and kornia loaders read; the upstream files
hold nothing else. The kornia maintainers converted them (converter revision 0c3db846b2). They checked that a
model loaded from either file gives bitwise-identical outputs.
| File | Upstream source | Source sha256 | sha256 |
|---|---|---|---|
RealESRGAN_x4plus.safetensors |
RealESRGAN_x4plus.pth |
4fa0d38905f75ac06eb49a7951b426670021be3018265fd191d2125df9d682f1 |
e82eee5fa7456f49025e70d906cd12aef0a90e5360d31d722488e9775156bce4 |
RealESRNet_x4plus.safetensors |
RealESRNet_x4plus.pth |
a820b9bde89a874d7599d545567308ce6c128fc8754a53208eda016d40aa81df |
4ac13640362bcb4e8e3b2d8e437a1eb089cf61777597b1b702eb4a348195e59b |
RealESRGAN_x4plus_anime_6B.safetensors |
RealESRGAN_x4plus_anime_6B.pth |
f872d837d3c90ed2e05227bed711af5671a6fd1c9f7d7e91c911a61f155e99da |
0e4b91575a84dcf019de3df1cfcb44a914c5a4fd713170f03ca672d31daace01 |
RealESRGAN_x2plus.safetensors |
RealESRGAN_x2plus.pth |
49fafd45f8fd7aa8d31ab2a22d14d91b536c34494a5cfe31eb5d89c2fa266abb |
89f045faba541d3dda3a6d5370d4ea6f588f8474cada0a1ee5d65397e669c2a9 |
License
BSD-3-Clause, Copyright (c) 2021, Xintao Wang, the licence of xinntao/Real-ESRGAN. See LICENSE,
copied from that repository.
Citation
@InProceedings{wang2021realesrgan,
author = {Xintao Wang and Liangbin Xie and Chao Dong and Ying Shan},
title = {Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data},
booktitle = {International Conference on Computer Vision Workshops (ICCVW)},
date = {2021}
}