My code is:
However it's really really slow. It's encoding at 2.38fps, which takes hours at 2 pass. Is sComb() causing TFM to think the frames are all combed and causes QTGMC to deinterlace them all? I usually get like 8-16fps.Code:sComb(ntsccombmask=true) D = QTGMC(Preset="Slower", FPSDivisor=2) TFM(D2V="VTS_01_1.d2v", slow=2, cthresh=4, mi=25, clip2=D) TDecimate(mode=1) FineDehalo(rx=2.0, ry=2.0, thmi=80, thma=128, thlimi=50, thlima=100, darkstr=1.0, brightstr=1.0, showmask=0, contra=0.0, excl=true)
+ Reply to Thread
Results 1 to 10 of 10
-
-
You can use display=true for TFM to see which frames are actually being deinterlaced (it will overlay info and indicate when deinterlace is applied), then adjust the thresholds and settings to fine tune
What is the CPU usage when running that encode? You can try MT to speedup processing if the script is the problem. Or there can be other issues like maybe other bottlenecks, slow encoding settings, hardware I/O issues . You can run avsmeter to help diagnose problems and that can help you tweak things -
Did you try the 'Display=True' addition to TFM as pdr suggested, to see the number of frames being deinterlaced? That CThresh setting is so low that QTGMC will be kicking in all the time and slowing that encode way down unnecessarily.
Is the field matching really so poor that the CThresh has to be so low? Or, alternatively, can you remove that CThresh parameter completely (the default is 9) and use Vinverse for the post-processing? -
I removed the sComb() line and the FPS went back to 7. So I think sComb is causing TFM to have QTGMC kick in. I have the sComb line before the IVTC line as real.finder suggested on Doom9. I will see if cthresh at 9 will make it go faster with sComb.
I'm using sComb to remove the rainbowing on my source.
EDIT: Set the cthresh to 9, now it's 3FPS. When everything is removed but sComb it's like 4FPS. So it's 100% sComb doing the the slowdown.Last edited by Vitality; 7th Mar 2019 at 17:38.
-
Not exactly. You said 2.38fps with your original script and 3fps with the CThresh changed to 9 and the script otherwise being the same. A roughly 25% improvement. If I understand what you wrote correctly. But, yes, having this sComb in the script accounts for most (75% or so) of the slowdown.
-
-
-
Last edited by Vitality; 7th Mar 2019 at 21:27.
Similar Threads
-
X265 2PASS SLOW quality WAS achieved by NVENC HEVC (Staxrip)
By Comparison in forum Video ConversionReplies: 37Last Post: 14th Jan 2022, 03:42 -
How to achieve the quality of X264&X265 2pass slow with NVENC HEVC?STAXRIP
By Comparison in forum Video ConversionReplies: 10Last Post: 10th Jun 2020, 15:39 -
staxrip gpu encoding error
By nixiejames in forum Video ConversionReplies: 2Last Post: 21st Nov 2016, 01:28 -
Avisynth Slow Encoding Bug when NOT using MT!
By VideoFanatic in forum Video ConversionReplies: 10Last Post: 4th Dec 2015, 08:02 -
which version of staxrip X64 support webm encoding?
By milad24 in forum Video ConversionReplies: 4Last Post: 22nd Jul 2015, 09:31