Y/C Separation 2D/3D could this be the only reason?

[DB83]

Now I do confess to know pip-squeek about these technical terms and, when I capture, if it works and looks right, to my eyes, then it's fine by me.

But here is the conundrum. Normally I 'capture' VHS through a ADVC300 using Ulead (Media Studio v7) capture program but transcode directly to Mpeg2 with this combination. No issues here at all.

The issue arises when I do a DV(type-1) 'capture'. No player would play the file. Windows Explorer would crash when I tried to select it for almost any reason and I could not import it into Pro-coder for more processing. In fact, I could only import it into Media Studio video editor, re-save the DV from there and that would then play and could be imported.

Now I was probably playing around with the ADVC settings and noticed that I had 3D y/c separation selected in the filter tab (I think the default is 2D).

So, I reset this to 2D and did another DV 'capture'. Now this time, the clip played fine and it imported fine and without a 're-sample' in the editor.

So it seems that the 3D y/c separation is the culprit but I am just curious as to why. I read that the 3D gives the best quality which is why I probably selected that in the first place. Yet I also read that this arrangement requires a 'digital frame buffer' whatever that is. So if some kind sir could answer these these questions as to why I can not use 3D and what a 'digital frame buffer' is and how I could get one.

[jagabo]

Analog composite video has the luma and chroma channels multiplexed onto a single wire. The comb filter separates those components back into individual signals again. y/c separation happens before the video is compressed by the DV encoder. A 3d comb filter simply does a better job than a 2d filter. There should be no difference in the DV data (except the better picture using the 3d filter). Unless there some kind of bug in the ADVC300. But we would have heard screams from many users if that was the case.

A 2d filter works in two dimensions: across the width of the video and across the height of the video. A 2d comb filter works with only a few scanlines at a time, so it only has to have enough memory to store a few scanlines. A 3d filter looks at multiple consecutive frames as well as across the width and height of each of those frames. So it has to store multiple frames in memory -- the frame buffers.

[DB83]

Jagabo,

Thnx for the reply. I read it a little earlier which got me thinking. If there is no issue with the filter then the issue must be with the software that created the video.

So I have done a few more 'captures' both with WinDV which I know all you guys swear by and a later version of Ulead Media Studio (V8). And in all cases I could quite successfully create the DV using both the 2D and 3D filters.

So it does indeed seem that the software was not handling the filter for DV (the mpeg2 transcodings seemed fine) which created the problem video although it could be edited.