Problems with DVD to KVCD with PAL Interlaced
 

Hi guys,

I have been converting quite a few of my DVD collection and have had no problems with them using the excellent MA script.

However i had one problem with trying to convert one of my movies namely '28 Days'. The problem lies i think is that all my other PAL films when put through DVD2AVI showed up as Progressive but this one movie was Interlaced.

The reason i say it was a problem is because all of my Progressive movies came out excellent when predicting using jorel's excellent new pridictions Here and i usually end up with a good KVCD @ 352 x 288 and usually end up with an MPEG of around 750Meg.

But with the interlaced movie the prediction went belly up as they say and i did what i always did set up ToK using jorel's method and the finished movie can out 850Meg and would only fit on a 90Min CD which i dont have.

I tried it twice and had the same problem here is my Tok txt:
The movie is about 15 seconds under 100 minutes and so i used 2 samples per minute in ToK then i tried it with 1 sample per minute and still the same result.

So (sorry for the long post) I was wondering if i should use Boulder's script for Interlaced source: Here
Maybe this will help with better predictions with Interlaced sources?

Please help and thanks in advance. ;)

So can any1 please shed some light on this please. :( :(

First:
Are you shure that this type of pal movie is interlaced???
No matter what DVD2AVI says.... Did you made a preview using just a simple import/resizing script to see if there are really comb-artefacts??
If yes you shouldn't deinterlace cause of quality loss and therefore use Boulders interlaced version of the MA routine. Or try one prediction without MA just to see.

Refering to prediction....
What Kind of prediction you choosed in TOK? "New faster prediction"?
Try to rise the number of samples and decrease the sample length so that the prediction won't take very much more time.
Did you also let CQmatic do a prediction ... just to compare?

Because thats the point ... maybe unfortunaly there have samples been used during the prediction which don't give a good over-all average of the movies treatment. Because I never had a prediction problems specified to interlaced sources!
But I had prediction problems like in cases of movies explained above!
That there have been choosed wrong samples to give CQmatic or Tok the right average.

And thats what Im thinking about instead of a 2pass/CQprediction way:
A modification in CQmatic which sets an offset of maybe 2mins every second prediction turn, to give more "different" samples of the movie. After the a/b prediction routines there could be calculated an average of the resulted a/b prediction routine samples .... that would even give a more preciser result. :?:

Make sure that the movie actually is interlaced. Make a script in which you only load the .d2v file and load it in VirtualDub. If you see combing in scenes with motion, the movie is interlaced. If you don't see combing, it's progressive.

Do not trust DVD2AVI, trust your own eyes! :wink:

If you want to use the MA portion of the optimal script and you do have an interlaced source, you must use the script I gave in that thread. Otherwise you'll just ruin the video!

Thanks for your replies guys much appreciated.

@incredible

As stated in my opening post i used jore's new calculating method in Kwag's post: "New highly accurate prediction method reported by jorel!" The link to this is in my post above.

Ive had no probs using this prediction method until i use this 1 movie. :(

@Boulder

Do i just use a simple avs script with the MpegSource line and put my d2v project in there?

Maybe your good self or incredible would be so kind as to post a little test script here that i could use and process though VDub i would be most greatful. :)

Thanks again guys.

1. Yes I saw the lines refering to the recommended sample length settings, but I also suggested to modify the settings in way I mentioned, cause this will give you more samples and by this a more preciser (but taking longer) prediction.

2. I didn't either had problems when using the settings in TOK we know, but as I mentioned also, It depends upon the movie and where the samples are selected by TOKs prediction so they could give you not enough real velaues about the movie's average contents to result in a precise CQ!

3. Its very easy to create such a script by yourself:
Open the d2v in moviestacker, apply your specific settings and save the .avs to your HD. Open it by using Vdub and watch the edges (hands, arms, objects) in high motion scenes to be shure if you got an interlaced source or not.

;-)

PS: I think if your source movie is anamorph it can't be interlaced. :wink:
Cause if you would watch it Letterboxed on TV (The Player just decreases the size of the height and adds borders) the architecture of the interlaced lines would be destroyed because they're based on the movies height :D

Thank you very much incredible for the advice you have given me here. Yes indeed my movie is letterboxed so i think you may be right i now know thanks to you and boulder that DVD2AVI although a great program doesnt always get it right i shall use my standard script through MovieStacker and see how it comes out in VDub.

Thanks again. :) ;)