Video still has duplicates after decimation?

[embis2003]

I've recently tackled the task of restoring all of my VOB music video files so they look acceptable on high resolution screens.

However, its seems in nearly every telecined VOB I tackle, tfm().tdecimate() or even tfm().tdecimate(cycle=5) will always leave annoying duplicates in parts of the video. Sometimes there are so many that motion is actually smoother if I leave it undecimated. Or some cases, a cross-fade will have duplicates on one scene but not on the other.

Am I missing something really simple?

[johnmeyer]

You have to look at each video, before doing IVTC, to make sure you understand the cadence. TFM is a pretty good field matcher, but it sometimes needs to be tweaked, and TDecimate most definitely needs to have the Cycle/CycleR parameters adjusted for some videos.

What method do you use to determine whether there are duplicates? Remember, you often need to first deconstruct the video into fields, using a non-blended Bob or a Separatefields(), so that you truly understand the decimation pattern. Looking at the video by simply viewing frames will not show you what you need to know.

[manono]

An untouched VOB or M2V sample from a DVD might help. One that shows the problem after a simple IVTC (but without the actual IVTC).

[jagabo]

VOB music video files... nearly every telecined VOB I tackle, tfm().tdecimate() or even tfm().tdecimate(cycle=5) will always leave annoying duplicates in parts of the video... Am I missing something really simple?

Music videos often have shots sped up or slowed down to sync the video to the music. This is typically done after the film has been telecined (ie, as interlaced video). So you end up with duplicate frames (when shots are slowed) or missing frames (when shots are sped up).;