Of course, It's posted right there 528x480
And yes, download this: www.kvcd.net/video.en1 and put it in the Opt directory where you have ToK installed. Make sure you use ToK version 0.0.5.2

-kwag

Hey Vhelp,

The way you get your resolution setting is via your avisynth script. The size you set in there determines the sample encode size for ToK. So if you want 704x480 then load you source in Moviestacker and choose Xvcd and copy and past the Bicubic (or your preference) lines and the Borders or Letterbox lines to the appropriate spots in your script.

So if you want 528x480 then choose 3/4 DVD(Small), etc.

Hope this helps.

Racer99

@ Kwag...

Thanks a bunch PAL !! hehe..

I'm sure we'll get it :banghead: or else
-vhelp

I guess I'm too tired now, and I said "select in ToK 528x480" Sorry vhelp, thanks Racer

-kwag

hey Vhelp toks will just overwrite the video avs n the tok log the others you live alone, infact live everything alone man.

@ all..

>> hey Vhelp toks will just overwrite the video avs n the tok log the
>> others you live alone, infact live everything alone man.

Alright.., will do everyone PAL's..

I'm gonna do some playing around now
-vhelp

I'm now starting my "K19" encode (again ) at CQ=62.783.
Results in ~5 and a half hours
Just finished prediction on K-Pax and "The Boondock Saints", and it seems the CQ is correct, based on previous CQ values.

@Tenra,
The 10% prediction is being thrown way off by "WHO_KNOWS_WHAT"
Either AviSynth 2.52, the MA script, Sampler(), or a combination of all
It seems that right now we must stick to a full sampler cycle for prediction.

-kwag

I can't seem to find the link to ToK version 0.0.5.2


David

@ Kwag..

Have you at least tried the scenario I laid out earlier w/ respect to this ??

You know..
Perform an AVIsynth v2.08 and an AVIsynth v2.52 encode, but this time,
leave out the MA parts. REM them out, and perform your CQ test.

If v2.52 performs similary to v2.08, then at least you have your culprit..
MA, ..and you can begin working on tuning it.

What does that sound like ?? feasable ??
-vhelp

It's right here, DKruskie pal..

* http://www.kvcd.net/hedix/ToK_0.0.5.2.zip

Thanks vhelp


David

Kwag:
I had no problems with the Faster Tenra Prediction method for my encode of Chicago. This was an xVid, 01:48:52 long. Here were my settings and results:

Prediction Factor: 1.00
Length of Sample: auto (24)
1st group of passes: 1% precision, speed up by 10 times
Predicted CQ: 51.790
Required Video Size: 714,950,948
Final Video Size: 707,818,009
RESULT: 1% difference!

When I went back to full prediction with your modified settings, I got a predicted CQ of 51.250, which would have probably ended up in maybe a 2-3% difference. All this was done with your latest script, your video.en1 file, and ToK 0.0.5.2.

I'm going to try encoding a dvd rip of Superstar next and see what happens.

hi J-Wo..

just curious.. but, what resolution did you use in that test ??

I asked, because I noticed that low CQ of 51.x and your final size was..
* Predicted CQ: 51.790
* Required Video Size: 714,950,948
* Final Video Size: 707,818,009

Just wondering how does the picuter so far ?? good ??
As I was saying.. I ask because I want to start using 704x480 (or higher)
in my encodes, but I'd like to know from those peoples who have done a
few of these w/ ToK, let me know how well it came out, so I know what to
expect. AT least in the ones (test encodes) I've done, I was quite pleased
w/ the results !! when using 704x480.

Thanks for any input.
-vhelp

vhelp:
I'm using 528x480 resolution. I think ever since I changed to avs 2.52 and the MA filtering, I end up with LOW cq's around the 50s or 60s. But the picture is now BRILLIANT. I can't believe how much crisper it is, brighter, more vivid, with hardly any noticeable artifacts (at least on my dinky 23" tv). It's comparable to encodes I did back in the day of avs 2.07 (let's all reminisce, shall we?) with CQ in the mid to high 70's. Kwag or anyone, care to explain why that is?

I used the defaults like this. I'll keep on testing

It's all in the Motion Adaptive script, and the way the filters are used
It REALLY makes that much of a difference

-kwag

@audi2honda,

You're using 2% on that second pass
I think that's too high

-kwag

Kwag:
I'm not noticing a large difference in CQ between the older vs. newer prediction method. I'm encoding Superstar right now, and with the new faster prediction I get CQ 63.566. With the older method 0.5% precision, 63.509. And finally older method 1.0% precision, 63.441. I'll encode all three tonight (1h 34min encode time each!) and see which one came the closest. All three CQs were calculated using prediction factor of 1.0.

I'm doing one after another myself. We'll figure this out one way or another. This MA script makes movies look so good, I j ust want to make the most of my bits [/quote]

I'm with you 100%