Help w/ understanding theory with converting from vob to mp4, concepts, etc

[kmxe]

Hi,

I'm a video DJ trying to understand some concepts and best workflows for my situation. im not sure i speak the right lingo, and im not sure if my theory is sound but im hoping someone can help me with the basics.

Basically, most of my videos is sourced from DVDs. My required end file is an mp4 with keyframes every 15-25 frames (i currently edit videos and have one every 20 frames).

From reading various web posts and forum posts, there is a lot of talk about 2:3 pulldown, converting vob to an uncompressed format for editing, different fps, IVTC etc etc and im struggling to put it all together coherently.


So, based on my required end format and based on reading all these posts, i think the following is the ideal workflow to ensure cleanest, sharpest video for Video DJing (with questions):

1) Extract vob file from DVD

- Someone in one of these forums has suggested that most video is shot at 24 frames per second (or is it 23.976??), and when this is transferred to DVD, something is done to change it to 29.97 fps. Is this called 2:3 pulldown or something like that???

- If so, is the vob file that i extract from the DVD unprocessed in anyway? And to clarify, the resulting file runs at 24 fps (or is it 23.976)???

- Also, this seems to apply for most NTSC DVD's - what about PAL???

- Would the video at this stage be considered interlaced?

2) If I am looking to edit the vob file before converting to mp4, it seems the best thing to do before i edit in a video editing program is to

a) convert to an uncompressed file type
b) edit the video (including editing or replacing the audio component)
c) somehow re-combine the edited video and audio back into an uncompressed video file type

So somehow it only goes through one compression stage?

3) Once i have the uncompressed file, to convert to MP4 I also need to consider de-interlacing?

I've read that I should Inverse Telecine the video file at some point which from my understanding has something to do with when the video was original converted from the original 24fps to 29.97 fps, and removing the duplicate frames. I think this should be done before any actual video editing - is this correct?

I hope someone can help me get this understanding correct!

[manono]

If you're in Australia then aren't your DVDs PAL? If so, forget about 3:2 pulldown, IVTC, 29.97fps, and anything having to do with NTSC. It'll just confuse you and clutter your brain. It's difficult enough as it is.

[kmxe]

yes im in Australia but i still source a lot of dvds from the US for the music videos which arent released locally

[manono]

Well, then music videos are among the toughest of all to reencode properly. Some are shot on video, some on film, some are hybrids of both, some are converted from PAL to NTSC and vice-versa, along with other kinds of weird combinations. Maybe if you want to post some samples we can advise. If you've never done any of this before you have your work cut out for you.

[kmxe]

ive been editing videos from the past couple of years but i want to improve my workflow which should result in better finished files.

RIght now I have a PC based workflow (but will be moving to mac soon once i get FCP). What Im looking for is just an understanding to why I should take steps when, and what to look out for - in respects to how to do the steps ill look to work out later.

Right now I:

- Extract vob from DVD (using DVDDecrypter, but also starting to use MacTheRIpper)
- drag right into Vegas Video and edit
- Render to .mpg
- change extension to .vob
- using VisualHub convert to mp4 (de-interlaced), keyframes every 20 frames)

my mp4s are ok - but i think they can be much better. i want to learn all this and understand rather than stumble through a workflow which ultimately wont be the best practice. im already dreading having to go through all the ones ive edited!!

[manono]

Then learn some AviSynth. And there is no one-size-fits-all workflow. These things are so very different from each other that each video has to be treated on its own merits. Also, except for the DVD Decrypter stage, I don't do anything at all the way you do. Perhaps someone else would like to help. Good luck.

[kmxe]

i understand that each video has be treated differently but i need to first understand what i could potentially look for, why i would need to do certain steps, when i would need to do certain steps etc - im not looking for a single workflow for everything.

thanks anyways - ill check out avisynth

[Cogitation]

kmxe, I can tell you that one place you're losing a lot of quality is that "- Render to .mpg" stage. Mpeg2 is just not an efficient codec at all. Further, any (non-lossless) encode destroys picture information. What you're doing is like taking an mp3 file, reconstituting a wave file, then encoding to mp3 again, then reconstituting a wave file and encoding to Nero AAC/MP4 audio. Too many encodes are going to limit quality. Keep that dvd-decryptor step. That's pulling the data off the dvd. From that point it depends on what you need to do. You mentioned editing, so I assume you mean actually doing stuff to the video like adding fades or splicing, and not just using Vegas as a program to output a file that VisualHub will accept. Assuming you just want to encode the video stream on the dvd directly, you want to look at a single program that will do it for you. HDConvertToX. This will take your .vob's from dvd decryptor and encode them to mp4 directly in one step. This will minimize any quality loss from the DVD source, which is what I'm guessing you want.

[Cogitation]

By the way... HDConvertToX makes use of avisynth directly, without you having to feed it commands via command line (which is really quite a pain and complicated.)

[kmxe]

kmxe, I can tell you that one place you're losing a lot of quality is that "- Render to .mpg" stage. Mpeg2 is just not an efficient codec at all. Further, any (non-lossless) encode destroys picture information. What you're doing is like taking an mp3 file, reconstituting a wave file, then encoding to mp3 again, then reconstituting a wave file and encoding to Nero AAC/MP4 audio. Too many encodes are going to limit quality. Keep that dvd-decryptor step. That's pulling the data off the dvd. From that point it depends on what you need to do. You mentioned editing, so I assume you mean actually doing stuff to the video like adding fades or splicing, and not just using Vegas as a program to output a file that VisualHub will accept. Assuming you just want to encode the video stream on the dvd directly, you want to look at a single program that will do it for you. HDConvertToX. This will take your .vob's from dvd decryptor and encode them to mp4 directly in one step. This will minimize any quality loss from the DVD source, which is what I'm guessing you want.

ok thanks for that tip..... from the looks of whats available, looks as though .avi is the best option then?

Yes - im editing as in splicing, replacing the audio, cutting up the video and re-arranging and all that fun stuff.

[Cogitation]

I would say .avi is the most universally compatible option, but also not nearly the best. avi is just a container format. It holds the video and audio streams in a single file with additional sync information so they can be played back together. It's also a very old format that has fewer features.

I prefer using .mkv (check out mkvtoolnix) exclusively, as it not only allows easy lossless demuxing (separating video and audio streams from the file into separate files that can be edited separately) but allows lots of stuff like easy chapter files, subtitle streams and pretty much any kind of video or audio stream you want to put in it. You can put Real Media video with Flac audio in an mkv file and it'll play. It's the AK-47 of container formats.

mp4, which I mentioned before, is probably a second best option. It's a little less easy to work with in my opinion, but just as capable of being edited and spliced up as .mkv is. It might also be more compatible with a program like Vegas, (or it might be that Vegas will only work with mpeg2, in which case, you might want to consider looking at a different program that can edit video streams that aren't mpeg2. Because as I said, mpeg2 is just a very old, very inefficient codec compared to modern ones like h264, vc1 or even Xvid (which I think people need to move on from, but that's just me :P))

Remember, these are container formats, not the codecs used to encode the video or audio streams. For the video encode. It really depends on what you want. If you're just trying to cut them up and put them together to be uploaded to youtube, or for your own use, then I'd recommend using x264 (which HDConvertToX is a great GUI for.) If you're making these for someone else, then your needs will probably vary, but still the rule of minimizing as much as possible the number of encodes and using the most efficient codec available for the final encode is going to produce the best results.

Good luck.

You figure this kind of stuff out by experimenting anyway, so hopefully you don't mind fiddling with settings and checking out programs.