frame accurate MPEG2 editing question

[Special K]

I'm just trying to get an understanding.....is there software that will allow you to cut ANYWHERE you want, frame by frame in an MPEG2 file, without the need to re-encode the entire video? Maybe I've been using the term "key frame" wrong....I'm not sure.

Yes, the afforementioned "smart" editors all allow that. They only reencode cut GOPs. I use VideoRedo for this personally. As I understand it Solveig works similarly (though I've never used it personally). I know Sony Vegas has this ability too.

I'm a little confused now. If I make a cut using MPG2CUT2, specifically cutting off some of the end of the video, are you saying that MPG2CUT2 must reencode the entire video, whereas VideoReDo, Solveig, and Sony Vegas would not need to reencode the entire video?

No.

mpeg2cut cuts on GOP boundaries only. It's not frame accurate. It does not, cannot, re-encode

videoredo, solveigmm etc... when smart rendering, are frame accurate and will re-encode the few frames within the affected GOP. All other segments pass through untouched . GOP length for MPEG2 is typically around 12-15 frames. So those frames are re-encoded around every cut, unless the cut is right on the GOP boundary

So basically if I make a cut with MPG2CUT2 I stand to lose up to 12-15 frames because that is the GOP length and MPG2CUT2 only cuts on GOP boundaries and doesn't reencode anything, whereas if I use something like VideoRedo it can make the cut between frames and then reencodes the 12-15 frames within that GOP?

I just want to make sure I have that straight.

[Special K]

yes, cuttermaran expects audio&video raw streams.

But if I make a cut in the video stream, does cutterman make the corresponding cut to the audio stream so the two remain in sync? If not, then I don't see how this tool would be useful, because any cut you would make to the demuxed video stream would result in it being out of sync with respect to the audio.

Audio and video normally remain in sync after editing with Cuttermaran. There are only 2 things I know of which cause them to go out of sync: (a) errors in one or both of the streams, and (b) the audio delay in the audio stream header is wrong. Problem (b) is relatively easy to fix. Problem (a) is virtually impossible to fix and you'd have the same problem with most other MPEG-2 editors. VideoReDo is the only MPEG-2 editor I have tried that is good at dealing with errors in audio and video streams, and it isn't free.

I must be missing something in the replies, but my question is essentially this:

Let's say I have a 60 minute MPGE2 file. I demux it into a 60 minute M2V video file and a 60 minute AC3 audio file. I then use cutterman to chop the last minute off of the M2V file. Are you saying cutterman will automatically make the same adjustment to the AC3 audio file so that I will now have a 59 minute M2V file and a 59 minute AC3 file that can be remuxed and will still be in sync? If the same cut isn't made to the AC3 file as was made to the M2V file, then the video and audio won't line up anymore when they are remuxed. I would have to cut the AC3 file manually before remuxing it.

[poisondeathray]

So basically if I make a cut with MPG2CUT2 I stand to lose up to 12-15 frames because that is the GOP length and MPG2CUT2 only cuts on GOP boundaries and doesn't reencode anything,

Lose or gain. It depends on what side of the target frame you cut on, and where the target frame is in relationship to the GOP boundaries

whereas if I use something like VideoRedo it can make the cut between frames and then reencodes the 12-15 frames within that GOP?

It only re-encodes if it has to. If your desired spot to cut happens to be right on the GOP boundary, it doesn't re-encode anything. In that case, mpg2cut2 would work perfectly as well

[Special K]

So basically if I make a cut with MPG2CUT2 I stand to lose up to 12-15 frames because that is the GOP length and MPG2CUT2 only cuts on GOP boundaries and doesn't reencode anything,

Lose or gain. It depends on what side of the target frame you cut on, and where the target frame is in relationship to the GOP boundaries

You can *gain* frames using MPG2CUT2 to make a cut? What does it use as filler for the added frames given that it doesn't reencode anything?

whereas if I use something like VideoRedo it can make the cut between frames and then reencodes the 12-15 frames within that GOP?

It only re-encodes if it has to. If your desired spot to cut happens to be right on the GOP boundary, it doesn't re-encode anything. In that case, mpg2cut2 would work perfectly as well

Is there a way to tell within MPG2CUT2 and/or videoredo where the boundaries are, and/or the frames within the boundary?

[poisondeathray]

You can *gain* frames using MPG2CUT2 to make a cut? What does it use as filler for the added frames given that it doesn't reencode anything?

Let me clarify with an example

By "gain" I mean lets say you want to cut right beetween frames 8 to 9, keeping frames 0-8 for 9 frames total. But that's right in the middle of a GOP. Instead it will skip to the nearest GOP boundary, right before the next keyframe. So you're left with frames 0-14 for 15 frames instead of a 9 frame clip.

So you lose or gain. You might have extra frames, or too few frames compared to if you used a frame accurate cut. ie. the cuts won't necessarily be where you want them to be. You might some frames from the previous scene, or that scene might be cut short

Is there a way to tell within MPG2CUT2 and/or videoredo where the boundaries are, and/or the frames within the boundary?

For mpg2cut2 - If you navigate by keyframes, you can ensure the cut is on keyframe boundaries. Mouse over the buttons in the GUI, and a short description will pop up it's the "<" ">" buttons in the GUI. "Step forward or back 1 GOP"

[Cornucopia]

Well, I already wasn't impressed with the "professionalism" of the production company used, when they put filler in between (some) segments, but GEEZ! anybody who has been doing this for any time at all should know not to provide a CUT version of the audio stream.
I've done the same "transfer/capture->encode->split for shipping on smaller media" process for years, and it is SIMPLE to output frame accurate encodes to specific time/space boundaries and make the joining be a breeze - even with long-GOP material. And Audio is so small, it is inconsequential to give a complete, uncut copy instead of or in addition to cut versions.

IIWY, I'd not waste MY time on this and take it back to the production company and tell them to do it right or refund your money (or both).

Scott

<edit>And if they don't understand, or need assistance with getting it done right, they should own up to it. And talk to me, if need be.</edit>

[Special K]

Well, I already wasn't impressed with the "professionalism" of the production company used, when they put filler in between (some) segments, but GEEZ! anybody who has been doing this for any time at all should know not to provide a CUT version of the audio stream.
I've done the same "transfer/capture->encode->split for shipping on smaller media" process for years, and it is SIMPLE to output frame accurate encodes to specific time/space boundaries and make the joining be a breeze - even with long-GOP material. And Audio is so small, it is inconsequential to give a complete, uncut copy instead of or in addition to cut versions.

IIWY, I'd not waste MY time on this and take it back to the production company and tell them to do it right or refund your money (or both).

Scott

<edit>And if they don't understand, or need assistance with getting it done right, they should own up to it. And talk to me, if need be.</edit>

Let's see, they added a few seconds of blank blue screen at the end of the first DVD, right after they cut the video. I'm not sure why they did that. They required more than 1 DVD based on the length of the VHS tape I sent them, so they had to split it across 2 DVDs, which means they had to split the video somewhere. I suppose they could have put the entire thing on one dual layer DVD. You are correct they did not provide me with an uncut version of the audio.

Maybe videoredo will be able to cut out the few seconds of blank blue filler screen and seamlessly join the 2 video segments together. I haven't tried it yet.

I just emailed the service I used to see if they can redo it using a dual layer DVD, which would allow the entire uncut video to fit on one disc.