yukihoaa / waifu2x_snowshell Goto Github PK
View Code? Open in Web Editor NEWMultilingual waifu2x GUI shell for windows x64
License: MIT License
Multilingual waifu2x GUI shell for windows x64
License: MIT License
Hello! Maybe you can Ukraine version in future? I want and can help from this, and thanks for your great job!
If possible, could you add an auto-update feature?
Because I didn't know github much, I don't know that a new update was released, so I often keep using the old version for a long time when a new version is released. If you add auto-update, it will solve the problem for people like me.
It would be nice if it started automatically when the program started, checked for updates, and if a newer version was available, notified the user and asked want to update.
I would appreciate it if you would consider this.
Print all message in to error log file
For those of us without Intel chipsets, pretty much the only solution that doesn't take forever is https://github.com/nihui/waifu2x-ncnn-vulkan.
And I really love Snowshell GUI, so it would be awesome to have support for this
Describe the bug
It is not showing the progress table as it should. it started happening today when I used GPU (with TTA). Now it still happens without TTA
It was utilizing a 1.8GB Vram when it first happened, later on when I tried to just denoise them it was using 200 MB Vram and this is how I found that it was doing something in background, but it wasn't consuming power at all, I mean it told it is using very low power.
Desktop (please complete the following information):
[Edit: Added proof that it was working on background]
Windows Defender recognizes the main program of waifu2x - Snowshell v2.3 (waifu2x_snowshell.exe) as virus (Trojan:Win32/Glupteba!ml).
Using cudnn-10.2-windows10-x64-v7.6.5.32
version.
Error:
Log file created at: 2020/01/03 13:41:37
Running on machine: DELL7730-XXXXX
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
F0103 13:41:37.764295 3732 cudnn_pooling_layer.cpp:12] Check failed: status == CUDNN_STATUS_SUCCESS (1 vs. 0) CUDNN_STATUS_NOT_INITIALIZED
With default model there wasn't problem.
config.ini :
[Snowshell]
ConverterNum=1
Noise=0
Scale=4
GPU=1
TTA=0
Export=0
Confirm=1
Lang=1
Debug=0
ScaleRatio=5
OutputDirName=output
[Converter]
waifu2x-caffe=C:\Users\XXXXX\Desktop\Waifu2x\waifu2x-caffe\waifu2x-caffe-cui.exe
waifu2x-converter-cpp=C:\Users\XXXXX\Desktop\Waifu2x\waifu2x-converter\waifu2x-converter-cpp.exe
waifu2x-ncnn-vulkan=C:\Users\XXXXX\Desktop\Waifu2x\waifu2x-ncnn-vulkan\waifu2x-ncnn-vulkan.exe
[Model]
waifu2x-caffe= models\\cunet
waifu2x-converter-cpp=
waifu2x-ncnn-vulkan=
[CustomOption]
waifu2x-caffe=
waifu2x-converter-cpp=
waifu2x-ncnn-vulkan=
The file C:\Users...\waifu2x_snowshell.exe is infected with Gen:Variant.Razy.878564 and was moved to quarantin
Describe the bug
When I try to convert,
Failed to convert some files. Check "error.log"
and there is no information in error.log.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
Regular run of the application.
Screenshots
https://imgur.com/NFaOAsT
https://imgur.com/gFYJS1u
Hardware Info (please complete the following information):
Additional context
If I use 'Vulkan' or use 'Force CPU only', no error occurs. However, it is inconvenient because the 'Force CPU only' settings are not saved, and I want to know why this is happening.
As a result of checking, the GPU memory is not insufficient at all.
Describe the bug
This is probably a false positive, but the latest release (2.3) gets blocked by Windows Defender for containing a trojan. According to Virustotal no other antivirus engines detect it as such.
To Reproduce
Try to use it on Windows 10 with Windows Defender enabled.
Expected behavior
Regular run of the application.
Desktop (please complete the following information):
the original waifu2x website has a setting that changes the mode between "photo" and "artwork", im guessing each is optimised for the respective type of image.
however, snowshell does not seem to have this setting, unless i missed it.
so does it exist in snowshell? and if not, is there a way to enable it by editing some config files or using the CLI instead?
The Intel integrated graphics on my lowly dual core desktop Intel Haswell processor takes around 45 seconds to apply 2x-scaling and low noise reduction to a 960x544 image.
Interestingly though, setting snowshell to use my CPU results in the exact same process also taking around 45 seconds (albiet with my CPU overclocked to 4.6GHz), and task manager confirms that it's very much using my CPU instead.
I then got the idea to launch two instances of snowshell, one window set to use the GPU and the other set to use the CPU. Sure enough, it was able to then complete two 960x544 images with the same 2x scaling + low noise reduction in only around 50 seconds or so which is nearly half of the time it would take if snowshell tried processing the same two images only through the GPU or only through the CPU.
So wouldn't it make sense if snowshell had a built-in option to use both the CPU and the GPU when specifically converting multiple files? Now obviously it can't use the GPU and CPU at the same time for a single image, but this way would at least let it process two files at once rather than a single file at a time.
Also ideally it shouldn't just split the allocation evenly (that is, half goes to the GPU and half goes to the CPU) since that would be no good when using a GPU that is vastly faster than your CPU or vice-versa. Therefore it'd probably be best to just treat it as one single queue of files to convert for both the CPU and GPU and simply allocates the next image in the queue to whichever processor finishes first and is ready for another image.
...I did however notice that snowshell does give ever so slightly different results depending on whether the image is processed through the CPU or GPU, but this may only be a difference mathematically and not to a level that any human could actually notice since I discovered this difference via Paint.NET's "difference" layer properties setting rather than by actually looking at the results and the difference may be impossible to actually spot visually anyway.
Is your feature request related to a problem? Please describe.
Implementing a way to upscale an entire video through the Interface, instead of a picture.
Describe the solution you'd like
Drag and Drop Support for Videos.
Describe alternatives you've considered
I currently use a waifu2x implementation called "dandere2x", which is fast, but can result into blocky artifacting if the source material isn't on par with the block matching system of dandere2x. i would love to have a quick way to use the "real" waifu2x implementation for videos without CLI.
Additional context
Just wanted to say thanks for the software, its working pretty good for images :)
Since the default value of denoise level of waifu2x-ncnn-vulkan is 0, Noize reduce will be performed unless "-n -1" is added.
Snowshell doesn't know how to really work at all if you try to right-click an image file and then selecting Snowshell from "Open with..."
Initially it'll want to create a config.ini file, and then when trying to convert it'll fail saying to "check your Windows version".
As a reference, waifu2x-cpp GUI from v5.2 of that program works perfectly via the process of right-clicking an image and selecting waifu2x-cpp GUI from "Open with...".
I see the option in a number of implementations of Waifu2x, including Caffee and Metal. It would be nice to include 4x as a native upscale without needing to enter a custom value.
Hello, I tried to upscaled 702 images using waifu2x_snowshell but only 700 images were upscaled and 2 images were not upscaled. There is a problem that If lots of (at least 500) images are tried to be upscaled, part of images (around 0.5% of total images) are not upscaled always.
Please fix this problem. Thank you.
When closing the program with TTA set to enabled, config.ini will have TTA set to 1. However the TTA setting will be reverted to disabled upon restarting the program, and TTA will also be set to 0 after closing the program the second time.
Good evening!
I'm sorry but the snowshell is working only when I manually open files one by one. When I just drag a folder or a couple of files, nothing happens...
Also sometimes manual opening of a file doesn't work too (in random cases, and only restart of the app can help)...
What can I do to fix this problem? I have some folders with a few hundred images to resize, and that's too hard to open them one by one...
Thank you!
Now waifu2x-converter-cpp with GUI, I'm very happy! Because I use AMD graphics card.
But now it can only handle one picture, I hope you can add batch processing later.
Although I want to help you, but I am a designer, not a programmer.
And... I have translated it into Chinese, for myself.
In the last: forget my bad English.
I can't help but notice that the results I'm getting from snowshell are different than the results I'm getting with waifu2x.udp.jp.
One thing in particular I've also noticed is that snowshell v1.9 and v2.1 give identical results with the same settings (that is when using GPU for both), but waifu2x.udp.jp updated their algorithm in late 2018, and update that doesn't seem to be present in snowshell.
Does this mean that snowshell uses a different implementation of the waifu2x algorithm than waifu2x.udp.jp does, or has snowshell simply not had its waifu2x algorithm updated to a more recent version?
The two options under the "GPU" menu are "Use GPU" and "Use CPU".
...but a CPU is not at all a GPU.
But both a GPU and CPU are processors as that's what the 'P' means in both acronyms.
Therefore would it not be more accurate to use the label "Processor" rather than "GPU"?
This would also farther sense if/when the "Use both GPU and CPU" function gets added since that could then be a third option labeled something like, well, "Use both GPU and CPU". EDIT: ...or I guess not since you've now stated that this is not something that Snowshell will implement, so I guess we can scratch that off the list.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.