net2cn

Results 14 comments of net2cn

Actually I have the same confusion after understanding this algorithm. It seems that it works as a sharpening algorithm like USM but specially designed for anime instead of upscaling algorithm...

Hi, Actually you can. In this case, I've made a TamperMonkey script for the purpose based on the WebGL implementation in this repo. You can install TamperMonkey extension on browsers...

现在是自动检测屏幕刷新率然后根据刷新率设定FPS限制(比如我的是60Hz的显示器,就会锁30FPS) 不过由于这里的检测算法是软实现的,可能会有偏差。因为Webkit限制,只能纯JavaScript实现检测算法,没办法直接读取硬件信息所以只能用requestAnimationFrame采样500ms重绘的帧数来检查刷新率。这个方法有时候测出来会偏低,比如我这里有时候会测出来59或者57左右,然后再除以2就会不足30,然后就直接让它如果计算出来的帧率限制小于30就向上取到30了

可以看看F12开发者工具里面Console回报的刷新率确认一下锁定的帧率 Edit: ![image](https://user-images.githubusercontent.com/6072596/108953616-7a186580-76a6-11eb-833e-0c7e8f77d146.png)

我晚点再找个高刷屏测试一下吧,可能是刷新率检测算法的锅

有,我想直接嵌入到B站播放器的设置面板来着 无奈前端功力不到家所以一直没做(要是有dalao愿意帮忙欢迎提pr

[1.0.0.3_pre1](https://github.com/net2cn/Real-ESRGAN_GUI/releases/tag/1.0.0.3_pre1)中添加了保留alpha通道并使用BICUBIC算法对alpha通道重采样放大。但是由于BICUBIC算法的限制,低分辨率的(测试使用的256x)带alpha通道的图片放大后图像的透明边缘会变得模糊。未来会添加可选用传统重采样算法如Bilinear、BICUBIC、Lanczos等,或使用Real-ESRGAN处理alpha通道的选项。

😄不一定会有,没有Mac开发环境而且也不是很熟悉Mac开发 还有一点是UI用的是WPF糊的,而WPF目前还是Windows限定 但是onnx模型本身并不是平台限定的,如果有后来者想做的话研究一下怎么在Mac下面跑onnx模型的推断,然后包一层UI库就好

Hmmm, I can replicate the same bug. This looks like a DirectML bug to me. I'll look into this issue.

The issue seems to be related to the OnnxRuntime with the DirectML backend. I cannot get the bug reproduced on some machines but machines with both integrated graphics and discrete...