Hybrid Source? How to encode?

[Steel]

I've recently come across several hybrid source dvds.

Band of brothers
eagles - hell freezes over

How do you get the Big3 to do these properly?

I've tried selecting hybrid, but it didnt work, and autodetect didnt work either.

[HyperYagami]

http://www.doom9.org/ivtc-tut.htm

the sad truth is, you'll HAVE TO do it manually.

[wmansir]

You might want to check out DVD Rebuilder. It's the only method I know off that can recreate hybrid sources exactly. However it is still beta.

Back when I did BoB I think I just said to heck with it and IVTCed the whole thing. It made the video a little choppy, and it actually inspired me to write some AviSynth macros to help convert the scrolling credits to nice still frame credits, so they didn't get butchered by the IVTC. I posted my functions here (http://forum.doom9.org/showthread.php?s=&threadid=39195&highlight=credits), but the attachments seem to have disappeared.

[Angelus]

The latest beta of DIF4U (1.4.7 beta) is able to handle hybrid content. As far as I know it's done a pretty good job with some of the TV shows I've encoded (24 and Angel). One thing you could do is what HyperYagami said, and do it manually. You could change your avisynth scripts and open them up in Vdub or play them in media player classic and see if the results are any better. I know my version of DIF4U utilizes decomb and it's ways to deal with hybrid content. Check out the Decomb documentation to see more.

[Steel]

Ive tried using the autodetect and hybrid in the latest DIF4U, as i said in the original post. IT DOES NOT WORK.

[Angelus]

You could do what I mentioned in the second part of my post: manually edit the avs scripts using the appropriate settings for decomb. Neuron2 has a good readme for decomb explaining how to deal with hybrid scripts.
An example of a script DIF4U created for me for a hybrid source is:

import("C:\Tony\DVD Backup Programs\DoItFast4U\new.avs\addaudio.avs")
LoadPlugin("C:\DVD Backup Programs\DoItFast4U\new.avs\mpeg2dec3.dll")
Mpeg2Source("VTS__02_P01.P.4~3_1.d2v",idct=0)
LoadPlugin("C:\DVD Backup Programs\DoItFast4U\new.avs\decomb.dll")
Telecide(guide=1)
Decimate(mode=1,threshold=3)
AddAudio()
ConvertToYUY2(interlaced=true)

Try it and see if it works.

[HyperYagami]

Different people want different ways to treat hybrid. What works for me might not work for you. There's no "right" way to do it, all depends on you and that's why you'll have to do it manually.

Since the OP never mentioned his/her definition of "it does not work" I assume that's the case.

[Steel]

What I meant by "does not work" is autodetect and hybrid in DIF4U. I goes thru the process and makes like 60k m2v files from a 4 gig original. That isn't in anyones definition of working.

[HyperYagami]

Originally posted by Steel
I goes thru the process and makes like 60k m2v files from a 4 gig original.

do it manually. that's your only hope. don't not expect miracles.

btw next time please say specifically what the problem is. that statement is a whole lot more useful than just "it does not work".

[Trahald]

Originally posted by wmansir
You might want to check out DVD Rebuilder. It's the only method I know off that can recreate hybrid sources exactly. However it is still beta.



That would assume the source changes at cell changes.(at least that what i THINK you are assuming ;) ) otherwise (at least in cce mode) your stuck with encoding at 23.976 for mostly film (which kills the video) or encoding at 29.97i which causes you to encode the rff'd video (ie encoding frames that didnt really exist before) .. while that would get you something really close visually to the origonal.. it definately wouldnt be like the origional. even if the video flags are changed a bit during remux.. there still either will be something missing.. or too much there. (depending on the method used).. having said that, cell demux will usually get you closest to the goal

[Steel]

in band of brothers disc2, the change back and forth from NTSC to FILM is during same vod and cell id. No way to split it and encode each seperately. vobid1/cellid1 has both ntsc and film. There really is no way to do it right.

[Matthew]

In theory, if you can generate a list of in and out points for each segment that is film and each segment that is video, you could use these in/out points in CCE in conjunction with a avs/d2v for the whole video. Encode the segments separately and then merge after the fact (e.g. import into Maestro, place on timeline and then export to m2v).

So long as you can generate the in/out points easily and reliably (let me guess - this isn't possible), then it actually wouldn't be all that hard to use this method - I think. One downside of course is that you'd no doubt be using a low average bitrate, so if a certain short segment scene is high action, you'd ideally want to allocate it some higher level of bitrate.

Leave the video stream untouched and backup onto 2 discs is my view =)

[RB]

Originally posted by Trahald
That would assume the source changes at cell changes.(at least that what i THINK you are assuming ;) ) otherwise (at least in cce mode) your stuck with encoding at 23.976 for mostly film (which kills the video) or encoding at 29.97i which causes you to encode the rff'd video (ie encoding frames that didnt really exist before)Not necessarily :) wmansir is actually right, DVD-Rebuilder currently is the only CCE-based solution that will recreate hybrid content 100% accurate.

The DVD-RB "trick" for NTSC is to write a D2V file that always has force film on and only delivers the "raw frames" to the encoder. That is, RFF flags are not written to the D2V but recorded elsewhere (in a .FLG file). This way what gets sent into CCE is a 23.976 fps video stream of only the "real video frames". Then during remuxing, the frame rate codes and RFF flags (read from the .FLG file) are corrected where necessary. IOW a "smart pulldown during remuxing", pulldown is only applied where necessary.

This approach could also be used for the Big3, someone just needs to modify pulldown.exe. Neuron2 is currently releasing DVD2AVIdg/MPEG2DEC3dg v1.2.0.0 (http://forum.doom9.org/showthread.php?s=&threadid=75399&perpage=20&pagenumber=2). There's a new field operation "Raw Encoded Frames" which does very much what DVD-RB does (needs MPEG2DEC3dg 1.2.0.0 of course). Then basically pulldown.exe would have to be modified to parse the D2V and apply pulldown following the RFF flags in the D2V.

[Trahald]

Originally posted by RB
Not necessarily :) wmansir is actually right, DVD-Rebuilder currently is the only CCE-based solution that will recreate hybrid content 100% accurate.

The DVD-RB "trick" for NTSC is to write a D2V file that always has force film on and only delivers the "raw frames" to the encoder. That is, RFF flags are not written to the D2V but recorded elsewhere (in a .FLG file). This way what gets sent into CCE is a 23.976 fps video stream of only the "real video frames". Then during remuxing, the frame rate codes and RFF flags (read from the .FLG file) are corrected where necessary. IOW a "smart pulldown during remuxing", pulldown is only applied where necessary.

This approach could also be used for the Big3, someone just needs to modify pulldown.exe. Neuron2 is currently releasing DVD2AVIdg/MPEG2DEC3dg v1.2.0.0 (http://forum.doom9.org/showthread.php?s=&threadid=75399&perpage=20&pagenumber=2). There's a new field operation "Raw Encoded Frames" which does very much what DVD-RB does (needs MPEG2DEC3dg 1.2.0.0 of course). Then basically pulldown.exe would have to be modified to parse the D2V and apply pulldown following the RFF flags in the D2V.

yeah.. i tried it out before my first post ;)

i see... so even though it says 23.976.. its encoding all the frames(fields). i was using dvd2avi to verify only but wasnt checking the duration. i was wondering where the other 5 fps (from the video parts) where going.. but they were actually still there.. cool.. it computes now.

yeah.. neuron2's app will definately be something to look into.

[Trahald]

It looks like pulldown wont take too much work to get it do what we want. Im working on it now in my spare time.

[Steel]

Thing is though, when using the Big3, pulldown is only used after the video is re encoded by CCE, not before.

[Trahald]

Originally posted by Steel
Thing is though, when using the Big3, pulldown is only used after the video is re encoded by CCE, not before.

The plan is.. for hybrid content. a special version of dvd2avidg will be used to make a d2v that will output all the raw content of the origional. you encode based on this file. then the special pulldown will take the encoded file and set the flags to match the origional file.

the dvd2avidg is still in beta testing and the pulldown is being worked on now.

[stanjr]

What is the status of the new dvd2avi and pulldown? Will they be incorporated into new versions of the big 3?

[Trahald]

Im slow.. it will be a while.. the dvd2avi version is still being debuged so i wont even release until that is fairly stable. I wouldnt hold off doing any projects you had planned waiting for it.

[wfn1]

any news or updates on this new pulldown project? i for one am dying to see how it will turn out!

[Trahald]

im waiting for the feature in dgdecode.. no eta when it will be implemented. im considering something like what zeul is doing in numenu (combo of rb's tool and dvd2avidg) we'll see...

[MLS]

Hate to resurrect this, but curious if this pulldown tool is applicable post-DIF4U1.4.8, and if so if it is still in the works.

I know Trahald has so many projects now I feel bad for asking :).

/MLS

[Trahald]

hehe... the newer dgindex' have everything i need (the ALL FRAMES flag and all the info in the d2v) .. i have to consider it.. i never really have an issue with hybrids to motivate me sufficiently to do it... after the next version comes out.. that will propably be the thing i work on. the next version will have Fmalibus distributed bitrate implemented for segmented encodes and some gui tweaks (office xp look menus, etc)

[MLS]

Cool, thanks for all your hard work.

/MLS