Video, Image and GIF upscale/enlarge(Super-Resolution) and Video frame interpolation. Achieved with Waifu2x, SRMD, RealSR, Anime4K, RIFE, CAIN, DAIN and ACNet.

Waifu2x-Extension-GUI

📜 中文版说明文档 | ➡ Official telegram channel

Download Latest portable package for Windows x64 PCs with Intel/AMD/Nvidia GPUs.

📝 Full change log

Supported Languages: English, 简体中文, 繁體中文.

The beta build has a faster update cycle than the stable build, which allows you to experience the latest features of the software in advance.

Beta builds are more unstable than the stable builds because the beta builds have not been fully tested before release.

What is Waifu2x-Extension-GUI?

✨ Key features:

• ✅ Great compatibility: Multiple built-in engines, compatible with almost all modern Windows PCs.

• 📊 Flexible multi-thread setting: You can adjust the number of threads freely while providing a complete set of settings for each engine, Take advantage of all the capabilities of your PC.
• 🚀 🚀 🚀 Multi-GPU support: Multi-GPU support for ALL built-in engines.
• 🛠 Numerous customizable settings: You can automate your workflow by adjusting the many customizable settings in the software.
• 🤖 Smart settings: This software can automatically adjust some software settings based on your PC hardware information and the information of the files you want to process.
• 😉 Easy-to-use graphical user interface.
• 📨 Continual maintenance and updates.
Built-in super-resolution algorithms:
Waifu2x / SRMD / RealSR / Anime4K / ACNet

Built-in super-resolution engines:
Waifu2x-caffe / Waifu2x-converter / Waifu2x-ncnn-vulkan
SRMD-ncnn-vulkan / RealSR-ncnn-vulkan / Anime4KCPP / SRMD-CUDA

Built-in Video frame interpolation algorithms:
RIFE / CAIN / DAIN

Built-in Video frame interpolation engines:
rife-ncnn-vulkan / cain-ncnn-vulkan / dain-ncnn-vulkan


PayPal

除了扫码捐赠现金, 您也可以每天打开支付宝首页搜“682796694”领红包并使用, 我也可以获得收入.

https://www.patreon.com/aaronfeng

1. This software won't upload anything to the internet, so we won't collect any data from you, we don't even have a server.

2. This software will only connect to the internet when checking for new updates and update the QRCode on the Donate tab, which
will download two .ini files and two .jpg files from GitHub and Gitee. (Gitee is a copycat version of GitHub in The

In conclusion, we won't collect any data from you.


💝 Credits 💝 :

LASAGNA
• Copy Audio Stream/Copy Video Stream settings ignored

Video and audio is being converted despite the options to copy the audio and video streams being ticked.

The previous version worked fine, and I've had to go back to using that one for the time being.

• Selecting sleep when done no longer works

Selecting sleep when done no longer works. I get a box asking whether I want to sleep immediately, and it goes no further. I want to leave a project running and have the computer sleep when done automatically This no longer works..

• Frames are being skipped

Whenever i try upscaling a gif it makes some kind of weird double frames and it looks like its skipping some frames in the upscaled one, it isn't like the original, not sure what's the problem.

• [Feature request] Add Real-ESRGAN to the list of upscaler

Is your feature request related to a problem? Please describe. No, this is merely a request to add an upscaler.

Describe the solution you'd like https://github.com/xinntao/Real-ESRGAN In there, is a ncnn version of Real-ESRGAN, which in structure is very similar to RealSR, but has superior video quality. it would be really cool to have it in your software!

Describe alternatives you've considered Theres no alternative, just a request.

Additional context I have no further context.

• v0.461，视频质量劣化(前期图像处理)问题

我测试一段视频，对比v0.37有着一眼能看出的劣化，囧 我一开始以为是视频编码器问题

但是我重新折腾几次发现 v0.461这个，在处理视频的时候，拆分图像上，就已经和v0.37有着画质差别（比0.37更糊） 然后，用这个已经劣化的图片再放大，就是糊上加糊

我已经把v0.461和v0.37的最终视频，拆分图像，以及放大图像，原始视频，都单独复制出来并上传网盘了 其中 160，眼镜的脸，和肌肉那块

232，妹子的边缘

链接：https://pan.baidu.com/s/10kf3j7e2KpJGJpAW0tw9yg 提取码：cgow

请查收

软件设置参数 自定义分辨率3840x2160 保持宽高比 视频，降噪3 视频，内部线程3

压缩JPG和保存为JPG都未勾选

引擎默认，块大小200默认

硬件参数 8700k oc4.69 64g ram 1080TI

烦请排查问题所在 如果需要，我也可以提供更详细信息

• [BUG] subtitle format:pgs could causes output to be malformed

Software Version Latest BETA or Latest Stable? v3.95.23-Premium

Describe the bug enabling keep mkv subtitles feature causes output to be malformed/broken

in vlc

video stays black with vlc logo unless manually scrub to a different time.. when scrubbed back to start, image shows but is still and doesnt actually play images, just audio.

in windows media player

video starts without problem. but trying to play any timestamp where audio is missing, it would immediately skip to next part where audio exists.

in both vlc and windows media player

large portion of audio is missing in the center.

How to Reproduce the bug Steps to reproduce the bug:

sourcefile: 27 minute long mkv segment: 40 sec model: RealESRGAN animev3 tilesize: 400
2D custom resolution: 1920x1080 2x scale keep subtitles mkv nvidia x265 encoding nvidia decoder

1. enable keep mkv subtitle feature
2. upscale and render
3. play output video.

Expected behavior output is not malformed if keep subtitles is enabled

Files you were trying to process when encountered the BUG If you encountered the BUG when processing certain files, please send me some of that files so I can do some research.(You can zip it before send it to me)

The settings you were using when issue occured You can save your settings than zip and upload the settings.ini, or just take screenshots.

VideoConfiguration_[Kayoanime] Domestic Na Kanojo S01E01_mkv_Waifu2xEX.ini.md

Desktop (please complete the following information):

• OS: windows 11
• GPU Model RTX 3080Ti
• CPU Model ryzen9 7950x
• Amount of RAM: 64GB
• Amount of VRAM: 12GB
• Amount of Storage: 2TB

Additional context VLC player small 1 minute samples arent having this problem

• Converted images are black

I'm using version 0.52 beta, my video card is an RX570. When using Waifu2x-ncn-vulkan engine the result is black images (see screenshot). This did not happen in any previous version. To get around this I'm having to use the waifu2x-converter engine, which is much slower.

• [BUG] Unable to split video into pictures

Software Version: 3.92.01 Stable Release

Description of Issue: Waifu2x Extension GUI is unable to split a video into pictures while upscaling and inteprolating to 8K (a resolution of 7680 x 4320). It starts the upscaling process fine but around the ~50 minute mark, it randomly says there is an error splitting the video into pictures. Video is around 5 minutes long and running at 24FPS 1080P originally. Rendered in DaVinci Resolve in the .mp4 format using h.264.

How to Reproduce: Not sure if it only works with this video for some reason, but take any high bitrate video that's longer than ~6 or so minutes, upscale to 8K and it fails halfway through the upscale. Make the resolution 8K and enable frame interprolation (for 8K 48FPS).

Expected Behavior: The program successfully splits the video and moves on.

Screenshots: Console Log (you can see that it was working for 40 or so minutes) The video I was attempting to upscale to 8K 48FPS

File: https://mega.nz/file/q8cyEY5Z#fneW9bhDCUY40C_0LvQWnoXdDvZKyEUE5QmyjHF2_ek

Settings File: [Warning] .="Do not modify this file! It may cause the program to crash! If problems occur after the modification, delete this file and restart the program."

Specs:

• OS: Windows 11 21H2
• GPU Model: RTX 3070
• CPU Model: i5 12600K (not overclocked)
• Amount of RAM: 32GB
• Amount of VRAM: 8GB
• Amount of Storage: 500GB boot drive, 1TB project drive (where the content I'm upscaling is saved)

Additional Context: It should be mentioned I tried upscaling this video at least 3 times, with all 3 failing. I restarted my computer on my 3rd try, still no luck.

https://www.patreon.com/posts/53351638

https://www.patreon.com/posts/v3-70-11-beta-53564792

Software Version Latest Stable

Describe the bug When attempting to upscale a PSD, the program reaches the stage of extracting layer information before crashing. A folder is generated in the output directory with an empty results.txt file.

How to Reproduce the bug Steps to reproduce the bug:

1. Make sure that the GIMP .exe path and scripts folder path are correctly set before running, otherwise the program will display an error message before getting to the point where the crash occurs.
2. Load the provided test file into the processing queue.
3. Hit start and wait.

Expected behavior The program should go through the steps of processing the PSD instead of crashing.

Screenshots Video recording of an attempt to upscale the provided test file.

https://user-images.githubusercontent.com/24994538/124498455-e2d9c200-ddab-11eb-9110-7a979be1660b.mp4

Files you were trying to process when encountered the BUG

test.zip

Desktop (please complete the following information):

• OS: Windows 10 20H2
• GPU Model: AMD Radeon 5700 XT
• CPU Model: AMD Ryzen 7 3700X
• Amount of RAM: 32GB
• Amount of VRAM: 8GB
• Amount of Storage: 930GB (569GB free)

➡🔗 https://github.com/AaronFeng753/Waifu2x-Extension-GUI/issues/432#issuecomment-1327483718

*Edited by @AaronFeng753

I'm running the latest stable version, but I've had this issue with previous releases too. I assume it has something to do with the tile size, but I've tried changing it and it's still the same.

Whenever I use the RealESRGAN models, it makes the tiles so obvious that some parts are very noisy, and some over smoothened. This doesn't happen when I use RealESRGAN with other apps.

• Program freezes the first time it opens during compatibility test

• Program freezes the first time it opens during compatibility test;
• Even after several minutes the other commands remain grayed out;
• Closing the program and reopening solves the problem.

My current Specs Intel Core i5-9400F 16GB DIMM DDR4 clocked @ 2400 MHz GeForce GTX 1660 SUPER GAMING OC 6G

• [BUG] The new RealESRGAN model is not in the update archive

Software Version Latest BETA

Describe the bug As said in the title, the model is just not present in the update archive making it obviously non functional.

• [Feature request] Add Nvidia Maxine AI Video Upscaler

Nvidia has released a full SDK of their Maxine tech (also used in their Broadcast app and the up-coming browser video upscaling).

The SDK package includes pre-compiled binaries and models, so it should be easy to integrate with Waifu2x-Extension-GUI.

From my limited testing of real-life videos, the best quality is VideoEffectApp with SuperRes and mode 0, and the end result is slightly better than Waifu2x, but not quite as good as Real-ESRGAN. It is however at least 5x faster than other upscalers on my RTX3090!

(Pending more testing: comparison pics, actual timed durations, anime videos)

Describe the solution you'd like

Have Maxine as an additional upscaler, just like how we have ESRGAN.

Describe alternatives you've considered

The upscaler can be used standalone, if @AaronFeng753 decides not to implement it into Waifu2x-Extension-GUI.

SET PATH=%PATH%;..\external\opencv\bin;..\..\bin;
VideoEffectsApp.exe --effect=SuperRes --mode=0 --model_dir=..\..\bin\models --in_file=..\input\in.mp4 --resolution=1080 --out_file=out.mp4


Limitation: The video can only be scaled to 1.3333x, 1.5x, 2x, 3x or 4x times the input, no arbitrary resolution.

General info: https://developer.nvidia.com/maxine

Usage guide: https://docs.nvidia.com/deeplearning/maxine/vfx-sdk-system-guide/index.html

SDK guide but probably not relevant to us: https://docs.nvidia.com/deeplearning/maxine/vfx-sdk-programming-guide/index.html

• [Feature request] WaifuXL support

• Be civil and polite.
• Only BUG Report and Suggestions are allowed here, this is NOT customer-service center.
• You are only allowed to report bugs you encounter when using Latest version.(Beta/Stable)
• English and Chinese languages only.

Describe the solution you'd like Adding WaifuXL model inside waifu2x extension gui

Describe alternatives you've considered Well, its just that WaifuXL works much better than any model on waifu2x extension gui, so the alternative is just using the other models

Additional context I found out recently about this new model, WaifuXL, this is a Real-ESRGAN model trained on anime images, to be more precise, as the official github page says (where the model is, and yes, is open source) "For our super resolution network we are using the state-of-the-art Real-ESRGAN and for our image tagging network we are using a MobileNetV3. Both were trained on a subset of Danbooru2021." So, i did some tests myself and.. i can say that WaifuXL outperforms all the models inside (as of now) Waifu2x-Extension-GUI here are some examples on a photo completely custom: WaifuXL has sort of better "fine-detailing" compared to the standard RealESRGAN-NCNN-Vulkan, so, adding it is up to you

• [Feature request] Add Code Former (NeurIPS 2022) models

Please add Code Former's Real-ESRGAN models, as they are better at face restoration than vanilla Real-ESRGAN models. https://github.com/sczhou/CodeFormer

PoC black/white image sequence to dumpy gif image sequence converter

PoC black/white image sequence to dumpy gif image sequence converter

Dec 9, 2022
An image and texture viewer for tga, png, apng, exr, dds, gif, hdr, jpg, tif, ico, webp, and bmp files

An image and texture viewer for tga, png, apng, exr, dds, gif, hdr, jpg, tif, ico, webp, and bmp files. Uses Dear ImGui, OpenGL, and Tacent. Useful for game devs as it displays information like the presence of an alpha channel and querying specific pixels for their colour.

Dec 31, 2022
Import GIF/WebP animated image as a new AnimatedTexture asset type.

Animated Texture Plugin for Unreal Engine 5 This plugin allows you to import animated pictures into your Unreal Engine 5 project as a new AnimatedText

Jan 8, 2023
Video++, a C++14 high performance video and image processing library.

Video++ Video++ is a video and image processing library taking advantage of the C++14 standard to ease the writing of fast video and image processing

Dec 28, 2022

gif_load This is an ANSI C compatible animated GIF loader in a single header file of less than 300 lines of code (less than 200 without empty lines an

Dec 10, 2022
This program converts APNG animations into animated GIF format

apng2gif This project has been in disrepair for long time. Recently, there is need to change the number of loops in apng file. So I used it to modify

Oct 27, 2022
A GIF art engine that will allow you to generate multi-layer GIFs from single GIFs as layers.

A GIF art engine that will allow you to generate multi-layer GIFs from single GIFs as layers. All the code in this repository has been written by me in c++, inspired by the generative art engine of HashLips that does not support GIFs as layers. The problem arose from my and my teamleader's need to generate animated images and then GIFs, in the same way as HashLips generated static images.

Jan 2, 2023
The “Quite OK Image” format for fast, lossless image compression

The “Quite OK Image” format for fast, lossless image compression

Dec 30, 2022
Qt5 image viewer with optional video support

Qt5 image viewer with optional video support

Jan 1, 2023
C++ image processing and machine learning library with using of SIMD: SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, AVX, AVX2, AVX-512, VMX(Altivec) and VSX(Power7), NEON for ARM.

Introduction The Simd Library is a free open source image processing and machine learning library, designed for C and C++ programmers. It provides man

Jan 5, 2023
ppl.cv is a high-performance image processing library of openPPL supporting x86 and cuda platforms.

ppl.cv is a high-performance image processing library of openPPL supporting x86 and cuda platforms.

Dec 30, 2022
HDRView is a simple research-oriented image viewer with an emphasis on examining and comparing high-dynamic range (HDR) images

HDRView is a simple research-oriented high-dynamic range image viewer with an emphasis on examining and comparing images, and including minimalistic tonemapping capabilities. HDRView currently supports reading EXR, PNG, TGA, BMP, HDR, JPG, GIF, PNM, PFM, and PSD images and writing EXR, HDR, PNG, TGA, PPM, PFM, and BMP images.

Jan 5, 2023
The CImg Library is a small and open-source C++ toolkit for image processing

http://cimg.eu The CImg Library is a small and open-source C++ toolkit for image processing, designed with these properties in mind: CImg defines clas

Jan 3, 2023
Like feh, but better, faster, more image formats, simpler, more lightweight, animation support, and better UI

Like feh, but better, faster, more image formats, simpler, more lightweight, animation support, and better UI

Oct 23, 2022
This is a C++17 deployment of deep-learning based image inpainting algorithm on Windows10, using Libtorch, Opencv and Qt.

This is a desktop software for image inpainting. It is a C++ deployment of image inpainting algorithm on Windows10, based on C++17 and implemented using vs2019.

May 13, 2022
A fast image processing library with low memory needs.

libvips : an image processing library Introduction libvips is a demand-driven, horizontally threaded image processing library. Compared to similar lib

Nov 10, 2022
a generic C++ library for image analysis

VIGRA Computer Vision Library Copyright 1998-2013 by Ullrich Koethe This file is part of the VIGRA computer vision library. You may use,

Dec 30, 2022