View Full Version : Best way to convert NTSC video footage from DVD to XviD?
MaximRecoil
4th July 2005, 08:55
A lot of TV series that are being put on DVD these days were originally shot on video (NTSC 29.97 FPS interlaced) rather than film. When running these through DVD2AVI to make a D2V project file, all the guides I have read seem to indicate that you should set the field order to "none". Is there any particular reason why you shouldn't select "force film" to IVTC these to 23.976 FPS which would make for more efficient encoding to XviD?
Anyway, what can be done about the inevitable interlacing effects with these types of projects that will show up after they have been encoded? I haven't come across a guide that deals specifically with this and there seem to be forty-eleven different approaches to deinterlacing but usually only discuss a typical Hollywood movie that was originally shot on film and has been IVTC'd to 23.976 FPS.
Inventive Software
4th July 2005, 10:31
:search: :readguid:
There are many guides for this, and it's just like encoding normal content, only with a higher framerate. If the content is / was interlaced before transition to DVD, it's likely that either the content was seriously pre-processed before being encoded to DVD, or has many interlaced artefacts as a result of not being pre-processed. The only thing I could suggest is that you enable pre-processing if these artefacts are noticable and interfering with your viewing.
But if the source was originally shot in video 29.97 FPS, then encode it in 29.97 FPS.
Guest
4th July 2005, 13:33
@MaximRecoil
According to forum rule 12, there ain't no best around here, so please don't ask for it. Thank you.
:readrule:
communist
4th July 2005, 18:07
AFAIK:
IVTC is only for material originally shot on FILM (24 fps) that was pulled-down to 29.976. If your material is originally NOT from FILM but 'simple video' than you have 3 options:
1) encode it interlaced at 29.976 fps
2) deinterlace it to 29.976 progressive frames
3) deinterlace it to 59.94 progressive frames (Bob-deinterlace)
If you IVTC'ed 29.976 material that was originally shot at that frame-rate to 23.976 you'd get jerky motion.
MaximRecoil
4th July 2005, 18:47
AFAIK:
IVTC is only for material originally shot on FILM (24 fps) that was pulled-down to 29.976. If your material is originally NOT from FILM but 'simple video' than you have 3 options:
1) encode it interlaced at 29.976 fps
2) deinterlace it to 29.976 progressive frames
3) deinterlace it to 59.94 progressive frames (Bob-deinterlace)
If you IVTC'ed 29.976 material that was originally shot at that frame-rate to 23.976 you'd get jerky motion.
Option 2 sounds good, as option 1 is guaranteed to have the interlaced look during every motion scene and option 3 seems like it would result in a huge file if you wanted to get decent picture quality from it.
I tried IVTC'ing it and using the quick and dirty standard "deinterlace" (blend fields together) filter in VirtualDubMod and the results were not that bad. Jerkiness in the picture was only noticeable in a few scenes. The interlacing effects were gone but of course it left the ghost trail effect during motion scenes; which I suppose is better than the interlacing effects.
I found this from http://nickyguides.digital-digest.com/interlace-test.htm :
As a final note, the only truly 100% perfect deinterlace method I have seen seems to be the one that uses TMPGEnc. It is very hard to do but produces perfect results. My friend Doom9 has a guide on how to do it. Check him out Here.
Unfortunately, he only links to the main page of this site. Does anyone know which specific guide he is talking about? I searched this site for "deinterlace tmpgenc" and none of the 6 results looked right.
manono
6th July 2005, 10:49
Hi-
Forget Nicky Page's Guides. They were fine in their day, but are now, what, 4 or 5 years old? The TMPGEnc IVTC is obsolete (and very slow) when compared to the much faster and better AviSynth IVTC's available.
First you have to determine if your TV series is telecined film (2 of 5 frames interlaced), or was really shot on video (all frames interlaced). If Film, perform an IVTC. If video 30fps, then deinterlace (for AVI). If I were you, I'd first run one of the episodes through AutoGK which has a very good method of determining what the source is like, and how to handle it. When done, the log will tell you quite a bit about it.
If your TV Series DVDs are from recent TV series, the chances are good that they started life as Film. However, many are a mix of film and video, such as most of the Sci-Fi TV series, and are a real problem to handle.
A lot of TV series that are being put on DVD these days were originally shot on video (NTSC 29.97 FPS interlaced) rather than film.
Not true, at least not for the recent TV series.
Is there any particular reason why you shouldn't select "force film" to IVTC these to 23.976 FPS which would make for more efficient encoding to XviD?
Yes, because you'll get interlacing and jerky playback as a result when these things are encoded as Interlaced, but are really "hard telecined" Film. You'll have to IVTC them, if they are Film. You might get faster encoding, but certainly not more efficient encoding (whatever that means in this context).
MaximRecoil
9th July 2005, 21:01
A lot of TV series that are being put on DVD these days were originally shot on video (NTSC 29.97 FPS interlaced) rather than film.
Not true, at least not for the recent TV series.This TV series was a rather low budget one from the mid 80's. It was definitely shot on video. The show's original creator/producer has his own site and forum and I hung around there quite a bit while this DVD release was still a pipe dream and he had to do most of the legwork just to get released on DVD. It was shot on Betacam as was typical for 80's sitcoms, though the higher budget, usually hour-long shows from the 80's such as Magnum P.I. tended to be shot on film.
Anyway, here are a couple screenshots with information about the original VOB's:
http://maxim.skyphix.com/gspot.jpg
And, from DVD2AVI ('Field Operation' set to 'None'):
http://maxim.skyphix.com/dvd2avi.jpg
You mention AutoGK, does this have a deinterlace method or does it just determine what the source material is? I would really like to know of a good deinterlace method without going the 60 frames per second route. I have tried TMPGEnc's "Deinterlace (Even Field)" which completely took care of the interlace effects but seemed to introduce its own problems, such as artifacts associated with horizontal things on the screen, such as the roofline of a flat-roofed building; or horizontal stripes on a shirt.
manono
10th July 2005, 07:07
Hi-
You mention AutoGK, does this have a deinterlace method or does it just determine what the source material is?
It determines what the source is, and then if interlaced, as yours seems to be, it deinterlaces it using neuron2's own KernelDeint, making a deinterlaced, progressive, 29.97fps AVI out of it.
I have tried TMPGEnc's "Deinterlace (Even Field)" which completely took care of the interlace effects but seemed to introduce its own problems...
Yes, that's the equivalent of AviSynth's SeparateFields().SelectEven().LanczosResize(whatever,whatever). You lose half the resolution that way, and, as you noticed, it creates nasty artifacts, including aliasing/stairstepping and shimmering. No, there are much better AviSynth deinterlacers than that (although I think TMPGEnc has better ones also). My 2 favorites are KernelDeint and TDeint.
You might try running one of the episodes through AutoGK to see if you like the results. It will at least tell you the optimal settings for KernelDeint that you'll be able to use should you decide later to encode the series yourself.
MaximRecoil
10th July 2005, 20:41
You might try running one of the episodes through AutoGK to see if you like the results. It will at least tell you the optimal settings for KernelDeint that you'll be able to use should you decide later to encode the series yourself.
OK, that KernelDeint filter sounds good, I'd rather just try it in the routine I am already using. I only have a rough idea on how to use AviSynth filters, as I usually work with Hollywood movies that were shot on film and DVD2AVI's "force film" IVTC is good enough for me concerning that, and I resize and crop with VirtualDubMod filters. I do use AviSynth in the routine but only as a frameserver to get DVD2AVI's D2V project file into VDubMod. So if I could just implement the KernelDeint filter at that point it would be good.
This is the AVS script I am using right now for that:
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\mpeg2dec3.dll")
MPEG2Source("E:\episode1.d2v")
So, I'm guessing that I download the KernelDeint filter, Version 1.4.0 (or should I get the Leak KernelDeint Version 1.5.4?), from - http://neuron2.net/kerneldeint/kerneldeint.html - and then put it in the plugins directory of AviSynth, then add the "LoadPlugin" part for it...
If this is correct, then what do I add to the script to implement it? A line with common and/or default parameters would be good if you know of one. Thanks for the replies so far BTW.
manono
11th July 2005, 02:30
Hi-
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\mpeg2dec3.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\KernelDeint.dll")
MPEG2Source("E:\episode1.d2v")
KernelDeint(Order=1) #if TopFieldFirst, as most DVDs are
1. Read the docs to tweak, if you aren't pleased with the results.
2. Your DVD2AVI/MPEG2Dec3.dll are old. Get up to date DGIndex/DGDecode.dll from neuron2's site.
3. You should be using AviSynth Crop and Resize, rather than VDubMod's. This allows you to use Fast Recompress, rather than Full Processing, and the encoding will go significantly faster. One easy way to get the Crop/Resize values is to use Gordian Knot for this. It can also create your complete AviSynth script for you.
4. If the DLLs are in the AviSynth Plugins folder, they should load automatically when you open the script in VDubMod, so you don't really need to have them in the script. But before you go deleting those lines, you might make sure your scripts are working.
5. At the risk of being wrong, I'll say that I don't know that there's much difference between KernelDeint and LeakKernelDeint. Leak's may be a bit faster, but they're both pretty fast.
6. None of this has anything directly to do with XviD, so if you have further questions about AviSynth usage, maybe either post them in the AviSynth Forum, or PM me.
MaximRecoil
11th July 2005, 06:39
That is a great reply. I don't keep very up to date on this stuff because I don't do it very often anymore, since it is so simple to back up DVD's to a DVD-R these days. I am doing this project for portability reasons. I already have the season set on DVD, but my goal is to fit all 21 - 25 minute episodes, plus 1 - 29 minute episode, onto a single SL DVD-R. I am using 2 pass in an old version of XviD with 512x384 resolution and they are coming out pretty good, all but the interlacing problem. They are averaging a little over 1000 kbps bitrate for the video stream (I am using 96 kbps 48 kHz stereo MP3 for the audio) as shown in Gspot. I set it so the filesize comes to 230 MB for the first 29 minute episode and 202 MB for the rest of the just-under-25-minute episodes, for a total of 4472 MB (it will actually be a bit less because I calculated it assuming 25 minute episodes whereas the actual lengths range from 24 to 25 minutes for all but the 29 minute episode 1; so some of them will come out to 201 MB, because of the spaced saved on the shorter audio file).
I read Neuron2's page about DGIndex/DGDecode and I will definitely start using them in place of DVD2AVI/MPEG2DEC3.
About using AviSynth to crop and resize, I had no idea that you could then use fast recompress in VDubMod; but it makes sense and I am very glad to have found it out. I always used VDubMod's filters because they were so convenient and the quality was good. I'm guessing that the cropping values from VdubMod would work to put into the AVS script; i.e. "12" for both X1 and X2 offset, and "0" for both Y1 and Y2 offset. I find Gordian Knot and similar programs to be cumbersome, what with its forty-eleven programs that it wants to install, which I already have in different directories, and its massive non-resizable GUI (I run 800x600 due to a crappy monitor), and BeSweet audio processing (I use BeSweet for certain things, but VdubMod's audio processing is just fine for XviD AVI's IMO)...it just isn't for me. I prefer to do only the necessary steps manually.
With all of this updating, I suppose I should update XviD as well. I have been using Koepi's XviD-04102002-1 build for years with complete satisfaction. Is Koepi's XviD-1.0.3-20122004 a good, stable replacement?
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.