Problem with DVD-RB and QuEnc on ntsc film sources

[JDay]

This has been mentioned before but seems to have been overlooked. I see it as a major issue, preventing QuEnc from being usable, so I'm giving it it's own thread. I don't think this should be too difficult to fix. (but then again, what the hell do I know?)

When you encode progressive film sources using QuEnc and DVD-RB you will experience "jerkiness" on playback with MPC and certain standalones. This seems to be a problem with the prog_seq headers. It seems CCE sets this to interlaced when you encode a progressive clip, while QuEnc sets them as progressive (not entireley sure on this- that's how it appears, anyway. Is there a program that can be used to view this data?). These both play fine on their own but when the QuEnc clip is run through RB's pulldown, the problem occurs. The QuEnc clip will pulldown fine using pulldown.exe with default settings, and will pulldown with DVD-RB if I use pulldown to set the prog_seq to interlaced, leaving everything else unchanged. (I've also set the header to progressive on a CCE clip, which then exhibited the jerkiness, after RB's pulldown)

Previous discussion on this can be found in these threads:
http://forum.doom9.org/showthread.php?s=&threadid=75848
http://forum.doom9.org/showthread.php?s=&threadid=75923

[JDay]

Bump. jdobbs, could you please comment on this?

[jdobbs]

I'll write it down. But I'm really not sure what the issue is yet. I'll have to look at the headers and see what the difference is in the way CCE and QuEnc encodes progressive sequences.

[Trahald]

Dvdlab has a nice mpeg parser.. it will tell you pretty much every flag in an mpeg .. the pro beta has a 30 day trial

[jdobbs]

I have some code I wrote that does that. I'll run it against both encoding phase M2V files and look for differences.

[JDay]

Thanks, jdobbs. I'll sleep much better at night, knowing this is being looked at.

[JDay]

Jdobbs, have you done a comparison of the headers yet? I ran some clips through dvd-lab and some other apps but wasn't able to find any useful info (the amount of info that I can view is very limited with the available free/shareware tools).

[JDay]

[sigh]
*BUMP*

[jdobbs]

[DELETED] I realized I was answering the wrong question here...

[sigh right back at you]

[JDay]

Okay, one more bump. If this isn't addressed, I might make a utility to automatically flag the files as interlaced using pulldown.exe, as a work-around, and not bring it up again for a while.

[Trahald]

the pulldown.exe source is open and documented well and easily modifyable . it has code in it (but is disabled) to give alot of info about an mpeg stream. you can modify it to send to a file.

[dragongodz]

i will also add what i asked at the end of that second link since nobody replied there.

what WOULD be interesting is to know if any other programs require the seq_header flags aswell since this has only come up so far with DVD-RB. so do other authoring programs require it aswell or do they correct it themselves ?
this will let us know if we absolutly must fix it in avcodec or if we need to encourage jdobbs to in DVD-RB. :D

[jdobbs]

Well, I see what you're talking about. The sequence header generated by QuEnc set the progressive flag -- while CCE doesn't. The confusing part is: It seems like it should be set for a progressive stream... I can change it easy enough, I'm just not sure I should.

[dragongodz]

The sequence header generated by QuEnc set the progressive flag -- while CCE doesn't.
doesnt CCE have an option to set progressive flag in either video or quality part ? been a long time since i looked at a trial version but i thought it did. :)
if it does then maybe that would generate the same.

The confusing part is: It seems like it should be set for a progressive stream
exactly, it should. the only case this seems to be a problem is with pulldown/telecining.

I can change it easy enough, I'm just not sure I should
exactly why i asked for people to say if other programs(authoring) had the same problem or not. without more information we dont know which really needs the change. as yet such information has not been exactly overflowing. :eek:

[JDay]

I tested DVD-lab, and it did the pulldown fine, and the compiled dvd works ok in MPC, although I haven't tried it in my standalone. I may do that later, if anyon'es interested. My source was a short clip from "hellboy" that I'd been testing in RB, straight from the d2vavs folder.

[lamster]

Originally posted by jdobbs
Well, I see what you're talking about. The sequence header generated by QuEnc set the progressive flag -- while CCE doesn't. The confusing part is: It seems like it should be set for a progressive stream... I can change it easy enough, I'm just not sure I should.

Could you make it an option, so that those of us still experiencing problems can see if it fixes it for us?

Thanks...

[jdobbs]

I don't want to make it an option, because the tinkerers will set it and blame the software when it glitches (sorry folks, but this is experience talking) -- I want it to be set correctly. The question to all the smart people at doom9: "Is this flag normally set on or off for telecined progressive sources?" From the first entry in this thread I think it may be "off" -- which is easy to fix.

[jdobbs]

I'm going to make a change the sets the progressive flag off in the sequence header on NTSC progressive (telecined) encodes during rebuild.

[JDay]

I tested version 0.62 with QuEnc, and it worked great. Thank you so much!

[jdobbs]

I aim to please... I hope this makes it a little more useful to you.

[lamster]

0.62 fixed my problems as well.

Thanks again!