OK, I have now encoded the video with the second script and the size is 656 Mb compared to almost 1 gigabyte with just the template and no visual degradation of quality This is the real stuff

The first script I haven't tried yet because it seemded like it would take forever to decode.

I have an AMD 2400 overclocked to true 2200 Mhz so it's got to be a very difficult script I will try it later.

Also-the correct settings in TMPGenc are source-non interlaced and destionation progressive right?

Or should I leave source as interlaced and just encode as progressive?

Otherwise the results are great. I'm really happy with the results.
I'll be recommending it to all my friends. I told them about the forum but for noobs this always looks kinda complicated. I really appreciate your patience and time. Really great of you guys.

Congratulations!

The reason why the first script is slow is that it adapts to the motion in the video, blurring the parts with high motion more than the static ones. This may or may not improve compression, I personally use the RemoveDirt().TemporalSoften() approach for captures.

The next step for you would be to start playing with file size prediction (or start using MEncoder), for that I recommed that you go through the CQMatic forum.

I've never heard about Removedirt. How efficient it is comparing to deen for instance or something else more "known" ?

It doesn't improve compressibility as much as a simple Deen(), but the difference was very small (2-3%) when I tested it. What I like about it is that it's a temporal-only filter therefore keeping the details better than Deen. It's recommended to use a softener afterwards, hence the ultra light TemporalSoften(2,3,3,6,2).

Here's the official page:

http://www.removedirt.de.tf/

Just a uick question. Is it possible for the cropped video (those 16 pixels) to be stretched to completely full screen? So that there are no visible boarders on each side?

The TV set will expand it, don't bother.

That is why it's called 'TiVi'-overscan

Hi guys.

I've been using this script for months now and I am quite sattisfied-I left out the cutting of boarders (overscan takes care of it) because many people I give videos to watch them on monitors of PCs and it would be annoying.

The script has been serving me well but I was wondering if you guys figured out any new stuff in the meanwhile-any new recommendations?

Also is there a filter to make the colors a bit livelier. Compared to the original video, colors are a bit less vibrant in the KDVDs I get. Is there something to make them a bit more vibrant?

Thanks!

Probably the only new thing that has come up is RemoveGrain, which replaces UnDot nicely giving a compressibility gain as well. Try just RemoveGrain(), which defaults to mode=2. For more compression (and possibly some artifacts), try mode=8.

http://www.removegrain.de.tf (the website may load very slowly!)

OK. I will try it. On a small sample it already looks fine and size is smaller.

Thanks.

I never saw that : There is really no more reason to use Undot then

I have to test that :-p

The SSE2 version doesn't work completely, I think that modes 5-8 are ones that work. I use mainly mode 8 for my TV caps, it's a nice preprocessor for RemoveDirt().

EDIT: kassandro recommends mode=2 for all purposes, better compression but no artifacts