VideoHelp Forum
+ Reply to Thread
Page 2 of 4
FirstFirst 1 2 3 4 LastLast
Results 31 to 60 of 102
Thread
  1. Originally Posted by Downgraded286 View Post
    Well, I did give your suggestion a shot (Decomb - Bob @ FPS 59.94) it just still didn't come out right.
    Handbrake can't handle complex material like this. The only way is to use AviSynth.
    Quote Quote  
  2. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    Well, I've been messing with AviSynth and I've finally got it running like I need it to. My only problem is that VirtualDub isn't using all of my CPU now. I don't know why, but at this point I call it acceptable. At first I thought I was I/O limited (because at this point my source video is actually on a drive on another computer) but that seems not to be the case, as I tried moving one to the local machine as a test and it did the same thing. It topped out my CPU before, not sure why it's not now.

    Anyways, the main thing is that the road to the end result I need is now clear. I don't mind if it runs slow, but I'd at least like it to use all my CPU. If it runs slow already that's fine, but it's not even using all it can for some reason or another. But the main thing is it's getting there.
    Quote Quote  
  3. If you're using single threaded AviSynth and slow filters AviSynth will be the bottleneck. You need to run a multithreaded build of AviSynth get more CPU usage.

    http://forum.doom9.org/showthread.php?t=148782

    Then your script (which is wrong for film based sources) would have to change to something like:

    Code:
    SetMtMode(5,4) # mode, number of threads
    LoadPlugin ("mt_masktools-26.dll", "mvtools2.dll", "nnedi3.dll", "RemoveGrainSSE2.dll")
    FFmpegSource2("South Park S03E01.mkv")
    SetMtMode(2) # mode 2
    QTGMC( Preset="Slow" )
    SelectEven()
    In general, the lower the mode number the faster the processing. But many source filters require higher mode numbers. The number of threads (only specified in the first SetMtMode call) can vary depending on the complexity of the filtering. Usually between 1/2 to 2x the number of CPU threads.
    Last edited by jagabo; 19th Nov 2015 at 17:13.
    Quote Quote  
  4. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    This is what my script looks like:

    Code:
    SetMTMode(5, 4)
    FFmpegSource2("Z:\Media\South Park S01E01.mkv")
    SetMTMode(2)
    QTGMC(Preset="Slow", EdiThreads=2)
    SelectEven()
    TDecimate()
    Why is my script wrong for film sources?
    Quote Quote  
  5. Member Skiller's Avatar
    Join Date
    Oct 2013
    Location
    Germany
    Search PM
    I would highly suggest to stay away from SelectEven() – let's not forget the animation is at 24 fps (max), so the motion would be quite uneven after deinterlacing and throwing away every other frame down to 29.97 fps, but it's
    "standard 3:2 pulldown motion" as long as you deinterlace to 59.94 fps which is fine.
    By bob-deinterlacing to 59.94 we replicate what you would see when watching the DVD on a TV.


    Originally Posted by Downgraded286 View Post
    I don't mind if it runs slow, but I'd at least like it to use all my CPU. If it runs slow already that's fine, but it's not even using all it can for some reason or another.
    That's because it's not multi-threaded. You can use SEt's MT build of AviSynth. At least with QTGMC I have no luck with it however, it will glitch up a frame here and there as well as the audio , so it's not an option for me.


    Originally Posted by Downgraded286 View Post
    Why is my script wrong for film sources?
    Because film sources should not be deinterlaced but inverse telecined (ivtc). However this is not a clean film source, it's mixed.
    Last edited by Skiller; 19th Nov 2015 at 17:49.
    Quote Quote  
  6. Originally Posted by Downgraded286 View Post
    Why is my script wrong for film sources?
    Because film is 24 fps, not 30. The bulk of the video will be 24 fps. But some shots will be other frame rates because they were sped up or slowed down. Or from other material being overlaid.
    Quote Quote  
  7. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I'll keep your suggestion to remove SelectEven() in mind. However, at least now, such glitches would probably be very hard to pick up on, because it's South Park. It's already bad animation anyway.

    I am using the MT build of AviSynth. And it has to be running multithreaded, to a degree, because it's using about 40% cpu. If it was truly single threaded, it should show 25% max... I had it up to full load before, not sure what happened. Perhaps adding Decimate() slowed it down, for some reason?

    @Jagabo not sure what you mean? My script results in 24FPS (23.976) fps, currently.
    Quote Quote  
  8. Originally Posted by Downgraded286 View Post
    such glitches would probably be very hard to pick up on
    It will be obvious on most panning shots. You'll see six little jerks every second.
    Quote Quote  
  9. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I just had a thought. If I de-interlace (or de-telecine) from 29.97fps to 23.976fps, wouldn't the audio then be out of sync? How to fix this?
    Quote Quote  
  10. Originally Posted by Downgraded286 View Post
    If I de-interlace (or de-telecine) from 29.97fps to 23.976fps, wouldn't the audio then be out of sync?
    No. You'll wind up with 20% fewer frames and the framerate will be 20% lower. The video length remains the same. The original audio remains in synch.
    Quote Quote  
  11. Member Skiller's Avatar
    Join Date
    Oct 2013
    Location
    Germany
    Search PM
    By deinterlacing followed by deleting every other frame and then on top of that deleting some more frames to reach 23.976 you have certainly ruined the video (even if it's South Park and not all that obvious most of the time).

    Why not just deinterlace to 59.94 and leave it like that.


    You have to decide whether you deinterlace or ivtc. What you are currently doing is a mixture of both.
    Last edited by Skiller; 20th Nov 2015 at 05:47.
    Quote Quote  
  12. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I just noticed something about my script:

    Code:
    SetMTMode(5, 4)
    FFmpegSource2("South Park S07E01.mkv")
    SetMTMode(2)
    QTGMC(Preset="Draft", EdiThreads=2)
    SelectEven()
    Why is SetMTMode there twice? I more or less copied the script I found here, at the QTGMC page, and removed what I thought I didn't need and plugged in what I needed to. However I just noticed SetMTMode is there twice. Wouldn't the later value of (2) override the first value of (5, 4)? Isn't this a bit redundant?
    Quote Quote  
  13. Many source filters will not work properly at mode 2. So you set mode 5 before the source filter, then change to mode 2 after. So, in your script, ffmpegsource() is running in mode 5, the rest in mode 2.
    Quote Quote  
  14. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    Just ran into another problem.

    I have this DVD, 720x480 resolution of course, but with 16:9 aspect ratio. So you could call it a widescreen DVD I guess. After I run it through avisynth/qtgmc/virtualdub, it comes out 4:3 again. I do not like this. How can I preserve the aspect ratio?
    Quote Quote  
  15. Originally Posted by Downgraded286 View Post
    After I run it through avisynth/qtgmc/virtualdub, it comes out 4:3 again.
    No it doesn't. It comes out 3:2 (720/480=1.5 or 3:2). You either set a PAR/SAR during the encoding or you resize and reencode for square pixels, something like 720x404. Or, if later reencoding for DVD you set the encoder for 16:9 again.
    Quote Quote  
  16. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I've had other rips that showed 720x480 resolution, but they actually displayed something like 834x480. They were already progressive, so I didn't have to run them through QTGMC. This one, however, displays the 16:9 resolution, but is interlaced. When I run it through avisynth/virtualdub it loses this and goes back to 720x480.

    I'm not sure how to preserve this. On my 16:9 720x480 rips, they display 834x480, but even mediainfo shows them as 720x480 (albeit with 16:9 aspect ratio specified). I found a way to force it to something like 850x480 in virtualdub, but that's the actual storage size. I'd like to try to keep it the way it was before running it through avisynth. I believe the pixel aspect ratio is different?
    Quote Quote  
  17. Originally Posted by Downgraded286 View Post
    When I run it through avisynth/virtualdub it loses this and goes back to 720x480.
    It was always 720x480.

    Originally Posted by Downgraded286 View Post
    I'd like to try to keep it the way it was before running it through avisynth
    As I already mentioned, you do that in the encoder. This isn't something for AviSynth unless you want to resize it to square pixels. Since you're no longer using Handbrake (are you?) have you decided which encoder you'll be using?
    Quote Quote  
  18. The only thing that was lost in AviSynth is the flag that tells the encoder what the aspect ratio of the video is. All you have to do is tell the encoder yourself.
    Quote Quote  
  19. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I do still use Handbrake. First it goes through AviSynth/VirtaulDub, but then I lose audio, subtitles etc, so I replace them with the tracks from the original file with KMVToolNix. After that, I run it through Handbrake to get the h.264 .mp4 file.

    Not sure how I can reset the PAR flag to what it's supposed to be... that's my main issue with this ATM. I can forcibly resize it, but that way I'm storing a bigger image on disc. I don't want that, I want to store it in 720x480 but play with the 16:9 pixel aspect ratio.
    Quote Quote  
  20. Normally you can enter custom x264 settings in handbrake's x264 Extra Options box. Unfortunately, handbrake uses colon as a separator between options, and x264 uses a colon between the two sar values, for example "--sar=64:45". So handbrake doesn't parse it correctly. I don't know of any way to escape the color so there's apparently no way for you to specify your own sar in Handbrake. There are programs that can change sar flags in elementary h.264 streams. But as far as I know, they only let you change existing flags, not add flags to a stream that doesn't have them.

    So basically, I think you're out of luck with handbrake. Use another encoder. I use the command line version of x264, usually with a batch file. I just drop an AVS script onto the batch file.

    Code:
    start /b /low x264.exe --preset=slow --crf=18 --sar=12:11 --output %1.mkv %1
    Put that in a batch file. Put a copy of x264.exe in C:\Windows. Drag and drop an AVS script onto the batch file. You'll get an MKV file with just video. Use mkvtoolnix to add other streams.
    Quote Quote  
  21. Originally Posted by Downgraded286 View Post
    I do still use Handbrake. First it goes through AviSynth/VirtaulDub, but then I lose audio, subtitles etc, so I replace them with the tracks from the original file with KMVToolNix. After that, I run it through Handbrake to get the h.264 .mp4 file.
    Sounds kind of convoluted to me. An alternative is to open your AviSynth script in RipBot264, add the audio and the subs, and set your SAR in the box beside 'Profile'.
    Image Attached Thumbnails Click image for larger version

Name:	RipBot264SAR.jpg
Views:	221
Size:	81.7 KB
ID:	34782  

    Click image for larger version

Name:	Ripbot264Subtitles.jpg
Views:	197
Size:	87.5 KB
ID:	34783  

    Quote Quote  
  22. Member
    Join Date
    May 2014
    Location
    Memphis TN, US
    Search PM
    With NTSC 720x480 for DAR 16:9, the SAR is 40:33.
    NTSC 720x480, 4:3 DAR, SAR is 10:11.
    - My sister Ann's brother
    Quote Quote  
  23. Originally Posted by LMotlow View Post
    With NTSC 720x480 for DAR 16:9, the SAR is 40:33.
    NTSC 720x480, 4:3 DAR, SAR is 10:11.
    Those are for ITU analog video captures where the DAR is contained in a 704x480 portion of the 720x480 frame. According to the DVD/MPEG2 spec, where the DAR is the entire 720x480 frame, the SARs are 32:27 and 8:9, respectively.

    The difference between the two specs is usually ignored when ITU caps are put on DVD.
    Quote Quote  
  24. Member
    Join Date
    May 2014
    Location
    Memphis TN, US
    Search PM
    You go ahead and use those numbers, jagabo. Enjoy.
    - My sister Ann's brother
    Quote Quote  
  25. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I found a clunky way to do it in Handbrake. Maybe VidCoder would work better. Handbrake has a few things in their GUI that could use some work...

    Ripbot looks like a good alternative to the way I currently do it, however it doesn't seem to support network drives, for some odd reason.

    I ran into this issue today:
    https://forum.videohelp.com/images/imgfiles/taYTUMA.png

    This is after processing with QTGMC. Haven't noticed this before. What's all this about? Did it possibly need an IVTC instead of deinterlacing?
    Quote Quote  
  26. Originally Posted by Downgraded286 View Post
    This is after processing with QTGMC. Haven't noticed this before. What's all this about?
    QTGMC didn't create it, if that's what you're implying.
    Did it possibly need an IVTC instead of deinterlacing?
    If it's been telecined then you IVTC it. Would the blended frame have be there if you had IVTC'd instead? I don't know. Hard to tell from just a picture.
    Quote Quote  
  27. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I just tried using TIVTC's TFM filter. The result has issues as well.
    https://forum.videohelp.com/images/imgfiles/h0biaZx.png

    This is my script. I'm not as well versed in TIVTC as I am QTGMC (not that I'm a QTGMC expert either... but this is the first time I tried using TFM)

    Code:
    SetMTMode(5, 4)
    FFmpegSource2("Z:\Output\MakeMKV Output\South Park S18e01.mkv")
    SetMTMode(2)
    TFM()
    TDecimate()
    Quote Quote  
  28. Originally Posted by manono View Post
    Originally Posted by Downgraded286 View Post
    This is after processing with QTGMC. Haven't noticed this before. What's all this about?
    QTGMC didn't create it, if that's what you're implying.
    QTGMC does that occasionally with cartoons. You can sometimes avoid it by changing some of QTGMC's settings. But cartoons often have shots slowed down or sped up after being telecined -- using field blending which creates such artifacts too. You have to go back to the source to see whether it was in the source or caused by QTGMC.
    Quote Quote  
  29. Member
    Join Date
    May 2014
    Location
    United States
    Search Comp PM
    I've seen things like this with unprocessed interlaced video, but after I ran it through QTGMC it didn't seem to be there, until I hit season 18 of South Park.

    One difference I noticed is this seems to be the only series of files I've seen with field order bottom field first listed in MediaInfo. I'll try forcing assumebff and see what happens...

    Bah, same thing happened with QTGMC AssumeBFF(). I'll try it next with TFM.

    TFM AssumeBFF() didn't work out, either...
    Last edited by Downgraded286; 12th Dec 2015 at 21:35.
    Quote Quote  
  30. Originally Posted by jagabo View Post
    Originally Posted by manono View Post
    Originally Posted by Downgraded286 View Post
    This is after processing with QTGMC. Haven't noticed this before. What's all this about?
    QTGMC didn't create it, if that's what you're implying.
    QTGMC does that occasionally with cartoons.
    Because it uses nearby fields to create frames from fields then, yes, it's possible. 99% of the time the blending remaining after the QTGMC/SRestore process is because there were just too many blended fields to begin with and/or because the unblending process isn't perfect. If both fields of a frame are blended, then you'll get blends in the output. In the scenarios you mentioned, any remaining blends would be intentional. If you do wind up with blends created by QTGMC, then maybe switch to a deinterlacer/bobber that doesn't rely on nearby fields, one such as NNEDI3. I've used NNEDI3 as the bobber before when QTGMC was spreading around dirt and specks while bobbing a field-blended source.

    I think Downgraded286 has figured out his particular example can't be unblended (without frame interpolation, maybe, or some other sort of manual intervention). In this case, though, a picture doesn't tell the whole story and without a video sample we can't be positive.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!