Vegas vs HDVSplit m2t capture

Laurence wrote on 4/26/2007, 12:38 PM
I'm coming to believe that I should stop using HDVSplit and start using Vegas to capture m2t clips from my HDV camera. The reason is that the Vegas m2t capture seems to split the clips on existing Iframes rather than simply the point where the timecode changes. I know that the beginnings and ends of HDV clips captured with Vegas are often a couple of frames from the preceding or following clips.

Whatever it is, it seems like Vegas handles m2t clips as captured and split by Vegas better than ones aquired by HDVSplit. They seem to load faster on the timeline, don't stutter at the ends, and seem easier on the CPU.

I am also suspicious that the way HDVSplit separates clips might also be responsible for the dreaded "two black frame" bug that is driving some of us nuts.

Comments

mbryant wrote on 4/26/2007, 1:16 PM
I've been using Vegas to capture HDV, and have been regularly getting the "two black frame" bug.

Mark
Laurence wrote on 4/26/2007, 1:27 PM
Darn, I'm so sick of that problem.
goodtimej wrote on 4/26/2007, 1:49 PM
Wait, you get the 2 black frame thing also? I didn't know that was a component of Vegas capture!!! WTF!?!?

Is it really? What is known about it? What can be done? Should I be capturing with something else?
riredale wrote on 4/26/2007, 2:17 PM
Do a search on this board for the black frame issue. A lot of us have commented on it.

My assumption is that it's an internal Vegas issue, not related to capture.

As to HDVsplit, I would assume that timecode only changes on an I frame, because any new recording would have to begin with an I frame, right?
Laurence wrote on 4/26/2007, 3:00 PM
You can see the difference with a Vegas capture in that often you get a little bit of the previous scene at the beginning of a clip. This is because it is cut at the iframes and the iframes and timecode jumps don't always line up. HDVSplit always chopes exactly where the timecode changes regardless of whether or not there is an iframe there to begin with.

I'm pretty sure that HDVSplit chops the mpeg between the iframes and generates a new iframe and a bit of an out of sequence GOP at the front and back ends. I notice a little hesitation from Vegas at the in and out points and they are always cut exactly where the scene changes.

The reason I thought this might affect the two black frame problem is because I think that Vegas prefers an intact GOP sequence and works better when it has one. I thought this minimizing of strain might help out with this problem. I agree that there is no black frames upon capture either way.

In any case, I think that the "splitting at the iframe" approach that you get with a Vegas capture gives you clips that are easier for Vegas to work with because of this intact GOP sequence. My 3.06 P4 barely cutes it with HDV and needs every bit of efficiency it can get!

I really like capturing with HDV Split aside from this issue. It's use of GPU previewing makes it easy on the CPU even with a nice big preview window.
Laurence wrote on 4/26/2007, 3:02 PM
I also think the non-standard GOP sequence that HDV Split generates may also have been responsible for some of the problems I have had with Cineform files converted from m2t clips captured with HDV Split. I'm not sure yet, but it would make sense.
riredale wrote on 4/26/2007, 8:22 PM
HDVsplit is a freeware utility and I have no idea as to the technical competence of the author. That said, however, I don't see how ANY capture program would be able to split frames at anywhere but at an I frame, since the I frame itself is the foundation for the GOP--all the following P and B frames are just difference frames. Am I missing something here?

In any event, I have never used the Vegas m2t capture utility because my clips are organized around the date/time stamp that HDVsplit names them. So far everything seems to behave normally, but I'm certainly willing to entertain a switch to a different capture method of it is warranted.
Laurence wrote on 4/26/2007, 8:47 PM
Well you can split an mpeg anywhere if you rewrite the GOP sequence a little around the edit. Womble MPEG Wizard does this for instance. MPEG Wizard can edit on virtually any frame.

If you compare a Vegas m2t capture to a HDVsplit capture, you can see the difference right away. The HDVsplit capture splits right where the timecode changes. The Vegas capture splits are a little off from this. Often you'll see the last frame or two from the previous clip at the beginning.

My theory is that HDVsplit is rewriting the GOP sequence a little around the edit points and that that might matter to Vegas. I would love it if I was wrong and could go back to using HDVsplit. Aside from this issue it's a far better capture utility.
MRe wrote on 4/26/2007, 11:13 PM
Definitely new take must start with I-frame. Otherwise camera should store the information of previous I-frame and the changes thereafter somewhere. Or rewind to I-frame and recalculate everything before it starts recording new take. This would explain why it is so slow to start from cold. Anyway this is very unprobable.

Could it be so that the I-frame and respective timecode change are not 100% aligned? And HDVSplit then recalculates GOP sequence as you are suggesting.
Laurence wrote on 4/26/2007, 11:22 PM
I think it does it like Womble MPEG Edit: just stick in a new iframe, do a few frames, then jump to the next iframe already in the clip. This means that the first two iframes at the beginning of a clip are an unstandard distance apart. Womble does it pretty well but the out of sequence bits sometimes glitch a little. I think that Premier Pro must do something like this when it's editing m2t video in native mode as well.