Why does deinterlacing my u-matic 1 hour video take so long? Vdub2+Avisynth

[rcoltrane]

I don't know if my hardware is bad or if the script I'm using is slowing down the process, but when I try to deinterlace a 1 hour video using Vdub2 + Avisynth, it takes ages to process and I always end up aborting. I am using H264 10bit codec with lossless compression ratio. Here's the script that fixes chroma ghosting while deinterlacing:

LWLibavVideoSource("Tape 1 - AV Box 75.mp4", cache=false)
AssumeTFF()
SelectEven()
QTGMC()

even = SelectEven()
odd = SelectOdd()
interp = even.FrameRateConverter().SelectOdd()
odd = MergeChroma(odd, interp)
fixed = Interleave(even, odd)

return(fixed)

Please let me know if there's something I can do to speed up the process while keeping the best possible video quality, I mean without any compression loss.

Thanks!

[TubeBar]

I am not an expert in this area but why not use uncompressed video? That would be faster than h264 or any lossless compression.

[davexnet]

Your pc specs and the resolution of the video are contributing factors.
Check the cpu % in the task manager

[jagabo]

You should use AviSynth+ rather than AviSynth. Then enable multithreading by adding prefetch(N) at the end of the script:

LWLibavVideoSource("Tape 1 - AV Box 75.mp4", cache=false)
AssumeTFF()
SelectEven()
QTGMC()

even = SelectEven()
odd = SelectOdd()
interp = even.FrameRateConverter().SelectOdd()
odd = MergeChroma(odd, interp)
Interleave(even, odd)

prefetch(8) # ~number of cores/threads of your CPU.

[Alwyn]

prefetch(8) # ~number of cores/threads of your CPU.

The AVISynth wiki is confusing, because it says, for Prefectch:

default: (number of logical cores in the system) +1

There are, Logical Processors and then there are Cores, two different things.

Andrew, on his blog here, says that the Prefetch number is relates to the number of Logical Processors.

You can get the number of logical processors from your Task Manager CPU display:


[Attachment 72424 - Click to enlarge]

I have my Prefetch set on 20 and achieve around a 400% processing speed increase.

[jagabo]

And it's not even as simple as the number of cores or threads. It also depends on the particular script, the decoder(s), and encoder(s) you are using. You basically have to try a few different settings to get optimal results. Or just use a compromise setting that works for most scenarios. With my 8 core, 16 thread CPU I get optimal throughput somewhere Prefetch 8 and 16. Sometimes I have to go a little higher. And sometimes I don't want 100 percent cpu usage by AviSynth and the encoding (to leave CPU for other things, or just for quieter running) so I use fewer than "optimal".

[rcoltrane]

Thanks @Jagabo [this chroma script I'm using is yours ], @Alwyn and others for the quick replies!!

It seems that my laptop specs are not that good for this kind of thing. Here are my specs taken from Task Manager:


[Attachment 72637 - Click to enlarge]

I have just 4 Cores and 8 Logical Processors, with 8 gig of RAM. This seems to be the culprit here :/

Jagabo, I'm acutally using Avisynth+, sorry about that. I will insert that prefetch(8) line in my script as suggested to speed up things a little bit!!

Thanks again for your kind support on this!
Best,
RC

[pandy]

It seems that my laptop specs are not that good for this kind of thing. Here are my specs taken from Task Manager:

Your notebook may do throttling due insufficient cooling - this may be also important factor for declared slowness, side to CPU utilization verify also CPU temperatures and how quickly it raise from IDLE to full load. If temperature raise immediately after applying load to CPU and stabilize at relatively high value then it may be sign of the CPU overheat.