Episodic DVD to interlaced H.264?

[LocalH]

Hi,

I have a quandary, and I'd like to ask for advice from the forum. I have several NTSC DVDs that contain interlaced footage (a few are well-done professional PAL conversions like R1 Fawlty Towers, and a few episodes of the 80's game show Press Your Luck that I recorded from GSN when they were airing it). The Fawlty Towers discs are muxed into one large VOBset, with each episode in an individual PGC. The PYL discs I encoded as a separate title for each episode. I'd like to find an easy way to encode each episode to .mp4 and retain the interlacing (I absolutely despise seeing true interlaced material that's been deinterlaced, about as much as I loathe pan-and-scan or Stretch-o-vision™). Looking to aim for somewhere between 200-350MB per approx. 30 minute episode (Fawlty Towers will hit pretty close to 30min, being BBC and commercial-free, but the PYL eps sans breaks end up around 21-22 minutes or so). Does anyone have any suggestions for tools to use so that I don't have to do a bunch of manual work? I've looking into Handbrake, and it's awesome with the exception of a lack of interlaced support (at least, I saw no place to select interlaced encoding, nor field dominance).

[CarlEdman]

These days I generally use MakeMKV to extract DVD content before transcoding. This has two advantages in your situation.

First, frequently MakeMKV will cleanly separate episodes into their own MKVs which then can be transcoded separately.

Second, even if no separate titles are generated by MakeMKV, MKVs are easily chopped up and MakeMKV preserves chapter information. This allows you to chop multi-episode MKVs into single-episode MKVs while keeping video, audio, and subtitles in perfect sync. For an example of a python script which does this, see http://forum.doom9.org/showthread.php?t=162041.

Third, as for preserving interlacing, I'm not sure how to help. Almost all modern display technology (LCD, Plasma, LED, etc.) can't actually display interlaced content and has to deinterlace anyway. So unless you exclusively watch your content on a CRT, the only question is whether you deinterlace in a relatively unconstrained environment with a powerful processor at transcode or in realtime in the display device. If you choose a good filter, the former is almost always a better choice.

Originally Posted by LocalH

(I absolutely despise seeing true interlaced material that's been deinterlaced, about as much as I loathe pan-and-scan or Stretch-o-vision™).

Fawlty Towers isn't true interlaced material, though.

[hello_hello]

I generally use DVD Shrink to open the ripped DVD and then DVD Shrink's reauthor mode to resave it with each episode split into separate ifo/vob files. It's just a matter of dragging each title (episode) from DVD Shrink's right pane to the left pane then using the backup function. If the episodes are all one title, then you'd drag the title over multiple times (once for each episode) and use the edit function to set the beginning and end points accordingly for each episode. Of course you'll want to set the target output size to something large enough that DVD Shrink will just resave the DVD and not try to shrink it. Once you have each episode as a separate vob file it's fairly easy to re-encode them.

I don't know anything about interlaced encoding but if you're going to try to get old 4:3 video down to around 300MB per episode you're probably going to have to drop the resolution and/or quality a bit, and I'd suspect that's going to do more harm than deinterlacing will. Not having encoded video without deinterlacing I don't know how much it effects file size, but from the x264 Wiki:
http://mewiki.project357.com/wiki/X264_Settings#tff
"x264's interlaced encoding uses MBAFF, and is inherently less efficient than progressive encoding. For that reason, you should only encode interlaced if you intend to display the video on an interlaced display (or can't deinterlace the video before sending it to x264)."
So my logic would be to deinterlace in order to retain as much resolution/quality as possible.
Plus I'd wonder about device support for playing interlaced h264 video. It's not something I've thought about before.

I've compared interlaced video with the de-interlaced and encoded version in the past (briefly), and using the PC to deinterlace on playback the interlaced video tends to look marginally better in places, but it's a very small difference and to be honest I didn't experiment enough to discover whether those differences were due to the deinterlacing alone, or caused more by the resizing.

It's been quite a while since I encoded my PAL Faulty Towers DVDs but I think they were interlaced (I could be remembering wrong, I'd have to check them), however the NTSC version may not be. I used Xvid at the time so I can't tell you what file sizes to expect using x264.

Originally Posted by hello_hello

It's been quite a while since I encoded my PAL Faulty Towers DVDs but I think they were interlaced (I could be remembering wrong, I'd have to check them), however the NTSC version may not be.

Either way, Fawlty Towers was a mix of 16mm film and video. It was not true interlaced no matter how the video was ultimately mangled and encoded on the DVD. Now it may still be easier to just leave it interlaced if you don't want to go through the effort of trying to restore it.

[LocalH]

Originally Posted by amtm

Either way, Fawlty Towers was a mix of 16mm film and video. It was not true interlaced no matter how the video was ultimately mangled and encoded on the DVD. Now it may still be easier to just leave it interlaced if you don't want to go through the effort of trying to restore it.

Agreed, such as many programs of the day were. However, there are large sequences of true interlaced material (such as virtually all the scenes inside the hotel) and I want to preserve the motion fluidity of the same. Also remember - these Fawlty Towers eps are NTSC conversions, and they look to be done quite well with motion interpolation (camera pans are very smooth and general motion is virtually judder-less).

Originally Posted by CarlEdman

These days I generally use MakeMKV to extract DVD content before transcoding. This has two advantages in your situation.

First, frequently MakeMKV will cleanly separate episodes into their own MKVs which then can be transcoded separately.

Second, even if no separate titles are generated by MakeMKV, MKVs are easily chopped up and MakeMKV preserves chapter information. This allows you to chop multi-episode MKVs into single-episode MKVs while keeping video, audio, and subtitles in perfect sync. For an example of a python script which does this, see http://forum.doom9.org/showthread.php?t=162041.

Thanks for this suggestion, this seems to be a big step in the direction of what I'm looking for with regards to episodic DVDs. I shall be trying MakeMKV out shortly.

Originally Posted by CarlEdman

Third, as for preserving interlacing, I'm not sure how to help. Almost all modern display technology (LCD, Plasma, LED, etc.) can't actually display interlaced content and has to deinterlace anyway. So unless you exclusively watch your content on a CRT, the only question is whether you deinterlace in a relatively unconstrained environment with a powerful processor at transcode or in realtime in the display device. If you choose a good filter, the former is almost always a better choice.

Both of my PCs that I use to watch video do an excellent job of deinterlacing in hardware, and I prefer to delay the deinterlacing until playback as that allows one to make use of future deinterlacing algorithms that may be better than current ones (and I'd rather not deinterlace to 60fps as my computers drop a frame or two here and there playing 480p60 footage as compared to 480i60). Deinterlacing to 30fps and killing half the temporal resolution will not be happening :P

Originally Posted by hello_hello

I don't know anything about interlaced encoding but if you're going to try to get old 4:3 video down to around 300MB per episode you're probably going to have to drop the resolution and/or quality a bit, and I'd suspect that's going to do more harm than deinterlacing will. Not having encoded video without deinterlacing I don't know how much it effects file size, but from the x264 Wiki:
http://mewiki.project357.com/wiki/X264_Settings#tff
"x264's interlaced encoding uses MBAFF, and is inherently less efficient than progressive encoding. For that reason, you should only encode interlaced if you intend to display the video on an interlaced display (or can't deinterlace the video before sending it to x264)."
So my logic would be to deinterlace in order to retain as much resolution/quality as possible.
Plus I'd wonder about device support for playing interlaced h264 video. It's not something I've thought about before.

Hm, that's some interesting information regarding x264 interlaced encoding, thanks for that. I'm also not dead set on the filesize, I will gladly go as high as 700MB if that's what it takes to render acceptable quality without deinterlacing (although the smaller the better, given that this is old composite-sourced material I'll need more bits than similar footage produced with higher quality video interconnects). Given this, I'm also heavily considering switching back to ASP, as I've successfully made one-off interlaced encodes in the past from material I've recorded from television, and they've played back fine on every device I've tried them on that can successfully play back other forms of interlaced content. Still don't plan on using the old decrepit AVI container though :P

Thanks for the advice. I request that this thread stay open in case anyone wants to further the general discussion of interlaced encoding with x264 (or indeed other AVC codecs).