#qtgmc

wanted to see if there was any visible difference between the different qtgmc modes for gif making with avisynth.

my steps:

60fps H264 .ts file clipped into smaller .ts files (no file conversion or re-encoding, thus no quality loss... hopefully. I’m not an expert)

avisynth’s video folder > x264losslesswebm.bat (command prompt showed a bunch of errors but put the file thru anyway so I’m not exploring these errors. normalwebm.bat should work too)

deblinear, no extra sharpening

lighting/coloring psd and 0.03 frame delay applied to all gifs

conclusions:

to the naked eye these gifs look the damn same lmao

fast/slow doesn’t affect the flow of the gif

between 30 fast and 60 fast, there is no visible difference in the pixels of the frames. no difference in the frames of 30 slow compared to 60 slow either

HOWEVER... the pixels of fast are slightly less defined than the pixels of slow (see yonghee’s mole; 1000% zoom)

60 slow took So Fucking Long so i’m probably sticking to 30 slow

什么是交错，什么是反交错？简单来说就是因为视频制式的不同，出现了一些拉丝横纹，或者不正常的鬼影，比较影响观看体验，比较详细的可以搜“反交错”，知乎有个写的蛮详细的虽然不欢迎一切转载。

megui在创建avs脚本的里面有analysis交错的选项，分析以后给的脚本是

理应是可以解决手头视频的交错问题的，但是预览不了，翻了下16年的更新文档……发现需要“planar formats ”

Chikuzen18th March 2016, 15:54yadifmod2 = Yadif + yadifmod

latest version: 0.0.4 requirements: - Avisynth2.6.0 or Avisynth+ - Windows Vista sp2 or later - Visual C++ Redistributable Packages for Visual Studio 2015. github page (https://github.com/chikuzen/yadifmod2) binary download (https://github.com/chikuzen/yadifmod2/releases) (both 32bit and 64bit are included.) features: - All planar formats are supported. (and YUY2 support is dropped) - SSE2/SSSE3/AVX2 optimised.(MMX stuffs are dropped) - 64bit support. - not avisynth c plugin. Thanks to Fizick for the bug report.

至今还是不知道怎么操作才能用上yadif……一直报错，转成YV12还是报错，因为很急也搜不到就放弃了。

感觉各路前辈写的avs都和我的不一样……

也不是很想用TDeint，因为换了几个模式都有点奇怪的模糊……

然后是因为A卡，所以用不了 dgdecnv，感觉用 dgdecnv的话可以解决蛮多问题，语句也很好写

好吧终极方案QTGMC。

http://avisynth.nl/index.php/QTGMC

折腾过程：

=========================================================

1 找到了解决方案

效果很不错的滤镜合集。

QTGMC(Preset="Slow", FPSDivisor=2,)

一开始是使用Fast就能有不错的效果，其实默认预设就是slow，写不写无所谓。如果不写 FPSDivisor=2那么帧数翻倍，但是时长就短了。

==========================================================

2  初步决定 

速度实在太慢，而且CPU也吃不满，3700X也就10%的利用率。有多线程模式，但是 http://avisynth.nl/index.php/QTGMC上写的是SetMTMode()，是avisynth的语句 ，所以需要改成 SetFilterMTMode()。

To make this distinction clear from AviSynth+, SetMTMode() is called SetFilterMTMode() in AviSynth+.

于是加上 SetFilterMTMode("QTGMC", 2)

EdiThreads建议是半数核心多1

QTGMC(Preset="Slower", FPSDivisor=2, EdiThreads=5,)

测试了一下有6fps左右蛮可以的。

=======================================================

3 下一步 稍微添加

trim了1000帧没什么问题，场序没问题，那就用不上AssumeTFF和 AssumeBFF了，

SourceMatch是为了恢复细节，这个就参考QTGMC的默认好了

QTGMC( Preset=XXXX, SourceMatch=1, Sharpness=XXX, TR2=X )             # Basic mode, fastest QTGMC( Preset=XXXX, SourceMatch=1, Sharpness=XXX, TR2=X, Lossless=2 ) # Good quality, efficient, minor residual combing QTGMC( Preset=XXXX, SourceMatch=2, Sharpness=XXX, TR2=X )             # Good quality, slower, no combing QTGMC( Preset=XXXX, SourceMatch=3, Sharpness=XXX, TR2=X, Lossless=2 ) # Best quality, slowest, very minor residual combing

其实改了几个参数实际上的测试肉眼看不出来，所以无所谓了……

所以最终的avs是这样的：

#暂时不用SetFilterMTMode("LSMASHVideoSource", 3) LoadPlugin("插件地址/LSMASHSource.dll") LSMASHVideoSource("视频地址") SetFilterMTMode("QTGMC", 2) QTGMC(Preset="Slower", FPSDivisor=2, EdiThreads=5, SourceMatch=1, Sharpness=0.8, Lossless=2, TR2=1) Prefetch(8) ConvertBits(8) ConvertToYV12()

#crop #resize #denoise

速度还可以……继续翻doom9和nmm继续学学……