Sections of video which jitter

[DLister]

Hi there! I'm encoding dv from a Canopus ADVC-100 to mpeg2 with tmpgenc, and I have noticed that a few sections of each converted video, usually when the camera is panning while there is contrary motion in the foreground, have regular and noticeable jitter. Are the instants when it hitches caused by the I-frames, or is something else going on, perhaps something with the Canopus codec?

Thanks!

[rcb]

The advc100 has no impact on what you are seeing. But, yout TMPGenc encoding recipe can.
try a 720 x 480 2 pass vbr setting with around 6000 bitrate on an edited short avu clip from your main capture avi file containing one of the jitter scenes. Author and burn to a rw so you can see set top player results. If you like results, redo the whole capture with those settings.

Additionally, you can repeat the test , changing motion search setting to a higher level. See TMPGenc helpfile for details.
Good luck

[vhelp]

sounds like you are caping cable/satellite or something.

Your "camera is panning" is a bit misleading there.. as if you got a dv cam,
but that U R using an advc - very confusiong anyways..

I think your issue is that you got a film (movie) that you advc'ed and are
encoding it, but you are seeing a jitter motion that is more noticable during
the panning of certain scenes.

If the above is correct, then I would imagine that you would benefit from
an IVTC process in your project. You are probably seeing a hickup in a
3:2 pattern, which leads me to beleave your source is film.

The best approach to takling Film sorce in your encoding project is to use
an AVIsynth script that incorporates some line of code to process your video
.avi file. The script would look something similar to:

LN - Script Code 
01   movieDir = "c:\movies\"
02     dllDir = "c:\avisynthDLLs\"
03     
04   LoadPlugin(  dllDir + "decomb.dll")
05    AVISource(movieDir + "mymovie.avi")
06    
07   Telecide(swap=false, post=true, guide=1)
08   Decimate(cycle=5)

Remember to remove the LN 's. I displayed them for demo'ing sake.
Lines 01 and 02 is just a set of varialbes to hold your dir locations
Lines 04 and 05 load in the IVTC filter, and .avi source file to work on, and
Lines 07 and 08 are the actual lines that do the IVTC part from 04/05.

* .AVI source --> AVIsynth --> TMPG --> author to vobs --> DVD disk

-vhelp

[DLister]

Sorry about the misleading terms there... you're right, I'm not the one doing the actual panning. And yes, the original video is a combination of video and film, and I'm only seeing this on the film part. I wish I'd thought of that, actually! I'll give AVIsynth a try; I may just have to live with it, though, since the files contain both.

Thanks lots!

[DLister]

I'm trying to convert old vhs tapes, which came off cable, and which contain shows that are composed of film with video overlaid on them simultaneously. Will inverse telecining cause problems with the video part of the image?

Thanks again!

[vhelp]

decomb is good filter for most people experienced w/ it. I'm not a guru
w/ it comes to this filter because I got lazy and comprimised w/ the results
it DID give me. Actually, due to the nature/content of my source, though
advc device used here, the majority of my sources didn't have to much if
at all glitches. Only because my source was true film, and not a mixture,
unless done correctly by the Studios to begin with.

Mix sources are difficult to deal with. I know, having delt w/ my share of them.

The script I presented above is a basic script at best. It will work for all
video source as long as it's true film (w/out mixtures) etc.
.
.
If you have such mixture content, then you have to fine-tune decomb's
parameters, and that is where I lack the knoweldge thereof.

If you're having trouble w/ the color space (ie, YUY2 vs. YV12) etc., then it
is because of the latest version of AVIsynth defaulting to YV12 (though it is
suppose to bump down to the initial color space of YUY2, if your source is
such, if I remember correctly) But, you could still suffer from this issue is
you end up using plugins (ie, decomb) because some plugins (or filters)
will require one type of color space (ie, YUV or YV12 etc) So, therefore you
have to use a

* ConvertToYUY2, or
* ConvertToYV12 etc

for each filter that requires a certain color space to operate in. That is why
it is best to stick to one colorspace (or AVIsynth version that only uses one
color space (ie, YUV or YUY2) and use on those filters for THAT colorspace.
The problem there, is that you MUST know the colorspace of those filters,
and bring ONLY those filters inside AVIsynth scripts. Example, if your source
captured is YUV (or is it YUY2 - - I get confused too) then make sure that
any filters/plugins you use follow the same color space of YUY2 etc.
.
.
Note, it is said to be better to use YV12 colorspace, but the advc does not
capture to this colorspace. At best, it is YUY2 (if I understand correctly)
So, what one would have to do is use a * ConvertToYV12 after their inital
AVIsource("myfile.avi") and use ONLY those filters for this colorspace to
work properly.

I know, I know.., all too confusing. But, the only way to know is to do, and
to do is to learn or some domb boloney like that :P

Good luck,
-vhelp

[DLister]

Ok, let me make sure I have this straight. I'm going to run an inverse telecine process on my avi, thus making in progressive rather than interlaced at 24 fps. After I've done this, when I run it through TMPGEnc, do I use the 3:2 pulldown filter to get it back to 30 fps? I'm guessing that somehow this will not cause the artefacting to come back...

Thanks!

[vhelp]

yes.

In tmpg, set your fps to 23.976, and your other option to 3:2 pulldown.
-vhelp