Decombing

[FredThompson]

When DVD-RB is fed an non-pulldown NTSC film source and the "Deinterlace with DECOMB" option is enabled, what does that do? Does it yield a progressive result with the pulldown flag?

[jdobbs]

No. It uses FieldDeinterlace() to remove the combing effects. The purpose of the option is to clear up interlaced sources that have a combed appearance (even on the original).

[FredThompson]

Is there a method to convert to pulldown progressive within DVD-RB?

[jdobbs]

If the original is pulldown progressive, the output will be pulldown progressive as well. Hybrid input becomes hybrid on output also. But if you're asking whether DVD-RB can take an original that is not pulldown progressive (for example one that has been converted externally from 23.976 to 29.97 and then built as 29.97) the answer is "No". That's mainly because that situation is so rare -- and converting a true 29.97 source to 23.976 pulldown usually results in jumpy video.

I'll look at it as a possibility for v1.10.

[FredThompson]

That's exactly what I'm looking at, a film which was converted to 29.97 fps. It's easy enough to add Donald Graft's Decomb filter to the AVS for DVD-RB Pro. If I do that, will DVD-RB Pro know to set the pulldown flag for the resultant encodes or will there be other manual operations as well?

btw, This really is a fantastic tool, especially with the various support programs like RB-Opt and encoder farms.

[FredThompson]

Why do you say conversion back to 24 fps would be jumpy? An incorrect decombing would do that, not a proper one.

FWIW, I've tried decomb and decimate after phase 1 in each of the AVS files. DVD-RB Pro 1.01 pukes on that.

Source are Tora! Tora! Tora! and Heavy Metal. HM could really use some filtering. It's a horrible transfer. Looks like no cleanup was done, just a transfer from an old copy.

Suppose DVD-RB is used for phase 1 then each AVS is modified as needed to decomb and encoding done outside of DVD-RB then DVD-RB phase 3, the rebuilding, is done.

Where does DVD-RB Pro store encoder settings for each segment? Where is the rebuild framerate stored? (IOW, how to tell DVD-RB Pro to set as pulldown?) Would this mess up mux and navigation?

[jdobbs]

There's more to it than that. The flags have to be reapplied and the information DVD Rebuilder collected during PREPARE would all be incorrect... you'll have to wait for version 1.10...

[FredThompson]

Oh, heh, I see, there an INI for the encoder and also REBUILDER.INI such as this:

[Status]
mode=5
MovieOnly=0
Original_Size=3945436
Excluded_Audio_Sub_Size=0
MainVTS=07
VTS_05_SIZE=84875
VTS_07_SIZE=2887501
VTS_08_SIZE=478351
Progress=1
CCEType=3
[V0500001]
Last_Sector=87956
[V05000000001001]
SCR=.000
PTS=16642.000
Frame_Rate_Code=1
Pulldown=0
Structure=3
Frames=5324
Playback=6654
First_Sector=0
Last_Sector=87956
Reduction=50.5
Aspect_Ratio=2
HalfD1=0
Convert16=0
EndPTM=19998604.000
Audio_Sub_Sectors=2644
ILVU=0
Video_Sectors=42838

It looks like there is some elementary analysis because the values for Frames and Playback are darn close to 24 fps and 29.97 fps. These settings must be why the encoder (HC) puked. I modified the AviSynth script but the lengths wouldn't match after telecide and decimate because DVD-RB Pro expects the full 29.97 fps set of frames.

In truth, this looks like a task for RB-Opt which would keep you, jdobbs, free from questions from people who mess up. ("You used RB-Opt for which I don't give technical support...")

--

edit: I see what you mean. The D2V is all screwed up, too. Looks like a nice foundation you've got here, though.

Now the question you really dread...guesstimated delivery time for 1.10?

[jdobbs]

The D2V is screwed up? No, it's just tailored for serving raw frames.

[FredThompson]

I meant adding telecide and decimate to the AVS files messes them up. AviSynth will report problems with the D2V which, I assume, is related to the number of frames requested and the number available after "fixing" the video, similar to an out-of-bounds pointer. The statement should have been more explicit.

[TheSeeker]

There's more to it than that. The flags have to be reapplied and the information DVD Rebuilder collected during PREPARE would all be incorrect... you'll have to wait for version 1.10...

So your finally going to cave in and add IVTC support huh? Man people have been hounding you for that for ages it seems JDobbs. I can say it would be a welcome feature. Every now and again I get backups that have some pretty bad combing artifacts, that Im pretty sure could stand to be IVTC'd but I have never been able to do it with any sort of luck.

[jdobbs]

I decided it's worth it. I recently got some Simpsons Christmas discs.. (two of them), and they looked terrible on my player --- combing up the ying-yang -- and that was on the original DVDs. I looked at the structure -- and it was a telecined source that had been encoded as 29.97 interlaced. But, and here's the interesting part, they went even further and flagged the source as progressive.... so my progressive output player was, of course, slamming progressive (combed) images onto my screen.

Doing IVTC fixed the disc... but I had to do it by hand.

[TheSeeker]

I decided it's worth it. I recently got some Simpsons Christmas discs.. (two of them), and they looked terrible on my player --- combing up the ying-yang -- and that was on the original DVDs. I looked at the structure -- and it was a telecined source that had been encoded as 29.97 interlaced. But, and here's the interesting part, they went even further and flagged the source as progressive.... so my progressive output player was, of course, slamming progressive (combed) images onto my screen.

Doing IVTC fixed the disc... but I had to do it by hand.

Man what are some of these authoring houses thinking? Some guy must have come to work just totally hung over the day they were authoring those discs. Out of curiosity, how will IVTC work with DVD RB? I know there are quite a few different arguments that can be used (via the Telecide() function, which Im assuming is what your going to use), and the real trick is to know which to use and when. Will it be like if you enable ivtc, and then during the prepare phase the movie gets scanned and then at the end of the prepare phase the status log says something like IVTC arguments blah blah blah used, or it could say IVTC not needed are you sure you want to continue and use IVTC? Yes or No.

That would be the ideal way (for me, who has had nothing but issues doing ivtc). Also it would be a good way to find out if the dvd in fact NEEDS to have ivtc done at all. Would anything like that be remotely possible?

[jdobbs]

I haven't given it much thought yet. One thing you have to be careful of is hybrid sources (meaning the ones where you have a hybrid source encapsulated in a 29.97fps stream -- lots of TV captures have that).

My biggest fear is the people who have just enough knowledge to be dangerous. They will play a movie back on their PC and see combing (caused by the PC) and decide it needs to have IVTC and force it... only to have a terrible jumpy output. Then, of course, they will post here on DOOM9 telling the world that DVD-RB's IVTC code is crap. That's biggest reason it hasn't been added yet -- in fact some of the posts I've seen about IVTC requests have been concerning discs that I know aren't telecined.

[TheSeeker]

I haven't given it much thought yet. One thing you have to be careful of is hybrid sources.

My biggest fear is the people who have just enough knowledge to be dangerous. They will play a movie back on their PC and see combing (caused by the PC) and decide it needs to have IVTC and force it... only to have a terrible jumpy output. Then, of course, they will post here on DOOM9 telling the world that DVD-RB's IVTC code is crap. That's biggest reason it hasn't been added yet -- in fact some of the posts I've seen about IVTC requests have been concerning discs that I know aren't telecined.

See and thats just it. Im not quite knowledgable enough to know how to detect if a disc needs ivtc, although Im sure I could read up and figure it out eventually (actually I have a pretty good idea of how, its just doing the arguments for Telecide() that gets me every time). I think for the feature to be really anything more than a hidden feature, you would need some sort of detection scheme to say: Yes you need to do IVTC on this disc, or No this disc will not benefit from IVTC. Otherwise what you said will surely come to pass. Less then knowledgable users will come on complaining that their discs are jumpy and jerky and that everything is total crap. Im just not sure if implementing such a thing (the IVTC detection) would be worth the time it would take you to code.

Also, wouldn't you need some sort of audio stretcher/slowdown or something to change the audio length to match the new video length?

EDIT: And yes if your wondering I do have hidden motives behind my queries. I would just love to have a single, easy to use tool, that can detect and implement IVTC on a dvd that needs it, or tell me that it doesn't need it. It really is about the only thing that I can't do manually with any sucess.

[jdobbs]

You don't need to change the audio timing. When you remove the embedded telecined frames you change change the frame rate from 29.97 to 23.976, but when you add the pulldown flags it will push it back to 29.97 upon playback.

Trying to detect a hard telecined source is not that easy. You'd have to do some sort of frame-by-frame search for combing that fits the telecine pattern. I'll have to look around and see if there is an AVISYNTH filter that can do that. I know telecide()/decimate() can find them -- but if you have that command in it's too late.

[TheSeeker]

You don't need to change the audio timing. When you remove the embedded telecined frames you change change the frame rate from 29.97 to 23.976, but when you add the pulldown flags it will push it back to 29.97 upon playback.


Slaps forehead, duh. Your right, it gets flagged back up to 29.97 when it gets rebuilt.


Trying to detect a hard telecined source is not that easy. You'd have to do some sort of frame-by-frame search for combing that fits the telecine pattern. I'll have to look around and see if there is an AVISYNTH filter that can do that. I know telecide()/decimate() can find them -- but if you have that command in it's too late.

If figured it would be pretty difficult... You would almost have to have a small step added between prepare and encode (if IVTC is enabled), where it would take a decent sized cross section of segments made in the prepare phase. Add the proper functions and arguments to the avs files in question and run an encode logging the results to a file. Then that file would be scanned to see if IVTC is needed and pass those results on to dvd rb which would go back and change the avs files as needed. At which time the encode would run for real. The real question is would you have to run the whole movie through telecide/decimate to get a good idea if hard telecide is needed or would a fairly small chunk do the trick? Of course all of this is rhetorical, Im just sort of thinking aloud.

[manono]

AutoGK has a nearly flawless routine for detecting IVTC'able and hybrid sources. You might ask len0x about it. Either that or run a couple of hard telecined and hybrid sources through AutoGK yourself and then examine the .avs and log it creates.