Hugging Face's logo

aphrodite-tg
/
requel-models

Image-to-Image
ONNX
super-resolution
background-removal
non-commercial
Model card Files
xet
 Community

Requel models

Model weights used by Requel, a free, non-commercial image and video stylization tool.

None of these models were trained by the Requel authors. This repository is a mirror, kept so the application can download only the weights a user actually needs instead of shipping ~400 MB inside every build. Every file keeps its original licence, and the table below records the author and terms for each one. If you are an author listed here and would prefer this mirror not exist, please open a discussion on this repository and it will be removed.

Licensing, in short

This repository is mixed-licence, so the Hugging Face licence field reads other. Read the per-file terms before reusing anything.

Several files are non-commercial only. Requel is itself a free, non-commercial tool, which is the basis on which they are mirrored here. If you want any of these weights for commercial work you must obtain permission from the original author — for the AnimeGAN models the author explicitly offers a commercial authorisation letter on request.

Files

File Size Licence Author Original
u2net.onnx 168 MB Apache-2.0 Xuebin Qin and the U^2-Net authors source
RealESRGAN_x2plus.onnx 64 MB BSD-3-Clause Xintao Wang and the Real-ESRGAN authors source
RealESRGAN_x4plus.onnx 64 MB BSD-3-Clause Xintao Wang and the Real-ESRGAN authors source
UltraSharpV2.onnx 49 MB CC-BY-NC-SA-4.0 Kim2091 source
UltraSharpV2_Lite.onnx 29 MB CC-BY-NC-SA-4.0 Kim2091 source
AnimeGANv2.onnx 10 MB AnimeGAN-NC Asher Chan (TachibanaYoshino) source
AnimeGANv3_Ghibli_c1_e299.onnx 7 MB AnimeGAN-NC Asher Chan (TachibanaYoshino) source
AnimeGANv3_Hayao_36.onnx 4 MB AnimeGAN-NC Asher Chan (TachibanaYoshino) source
AnimeGANv3_Shinkai_37.onnx 4 MB AnimeGAN-NC Asher Chan (TachibanaYoshino) source

What each file is for

  • u2net.onnx — Background removal / salient object detection.
  • RealESRGAN_x2plus.onnx — 2x upscaling. ONNX conversion; upstream publishes .pth only.
  • RealESRGAN_x4plus.onnx — 4x upscaling. ONNX conversion; upstream publishes .pth only.
  • UltraSharpV2.onnx — 4x detail-preserving upscaling. Non-commercial use only.
  • UltraSharpV2_Lite.onnx — Faster UltraSharpV2 variant. Non-commercial use only.
  • AnimeGANv2.onnx — Anime stylization. Non-commercial use only.
  • AnimeGANv3_Ghibli_c1_e299.onnx — Ghibli-style stylization. Non-commercial use only.
  • AnimeGANv3_Hayao_36.onnx — Hayao-style stylization. Non-commercial use only.
  • AnimeGANv3_Shinkai_37.onnx — Shinkai-style stylization. Non-commercial use only.

Licence terms

Apache-2.0

Apache License 2.0. Redistribution and commercial use permitted; the licence text and attribution must travel with the file.

BSD-3-Clause

BSD 3-Clause. Redistribution and commercial use permitted; the copyright notice and disclaimer must travel with the file.

CC-BY-NC-SA-4.0

Creative Commons Attribution-NonCommercial-ShareAlike 4.0. Non-commercial use only. Attribution required, and adaptations must be shared under the same terms. https://creativecommons.org/licenses/by-nc-sa/4.0/

AnimeGAN-NC

AnimeGAN author's terms: freely available to academic and non-academic entities for non-commercial purposes such as academic research, teaching and scientific publications. Commercial use requires an authorisation letter from the author, offered on request by email.

Verifying a download

Every file's SHA-256 is recorded in checksums.sha256:

sha256sum -c checksums.sha256

Provenance

Where an identical file is published by the original project, its URL is given below. Files without one are conversions (for example ONNX exports of weights the upstream project publishes only as PyTorch .pth) and this mirror is the copy Requel uses.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Image-to-Image
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support