Avisynth: Latest Script Discussion - Page 13

girv · #**241** 05-01-2003, 04:57 AM

Quote:

Originally Posted by muaddib

Hi griv,

Just wanna give my 2 cents... because this script makes me really confused. :?

Quote:

Originally Posted by girv

Code:

LegalClip() 
LanczosResize(432,544) 
Blockbuster(method="dither",detail_min=1,detail_max=10,variance=0.3,seed=0) 
AddBorders(0,32,0,32) 
MSoften(3,3,4,3,pre=1,post=1,speed=4,showmask=false) 
Crop(0,32,432,544) 
MergeChroma(Blur(0.5)) 
MergeLuma(Blur(0.1)) 
LegalClip()

Look what you are doing:
First you resize to 432x544,
then you add 64 lines of top/bottom borders (32 lines each),
after that you crop all the borders you just added! 8O

So you end up with a final resolution of 432x544 <- that's a strange resolution!

I didn't get what you are trying to do... :roll:
I think that with this "strange" resolution TMPG will have to resize it too, and that's not good.

The borders are added because MSoften does not (seem to) process the very top and very bottom few lines of the image so you get left over DCT blocks flickering away like mad in those locations. Adding the borders gives the filter "room to work" at the edges, then I remove the borders afterwards as they are no longer needed.

As for the odd resolution, 432x544 is 480x576 with the overscan lines my TV does not display removed. I encode at 480x576 with TMPGEnc's "video arrange method" set to "centre" so it adds the borders required to bring the input clip up to the required resolution without resizing. Theres no particular reason for doing it this way, its just the way I like to do it.

girv · #**242** 05-01-2003, 05:01 AM

Quote:

Originally Posted by Wolfi

I tried your script with the same movie :? Kwag's script gave me a CQ value of 61.xx but your script was quite different, it gave me a CQ of 53.xx

Are you sure? Any source I've tried has come out smaller with my script at the same CQ...

jorel · #**243** 05-01-2003, 05:08 AM

Quote:

Originally Posted by Wolfi

3 nights

"the last script with STM filter is too slow but
testing the same sample with and without STM,
i see a big more details in scenes in the back of the movie.
little details are clean."

You say that you see more details with STM filter

//Wolfi

yes friend,believe me in some scenes i got more details
using the same sample with STM filter.
i don't know (how i could)but if is possible,
i can send pictures with and without the filter from my samples.

ps:
i sleep one 2 or 3 days in the week.
and my friends(some are doctors)stay

Wolfi · #**244** 05-01-2003, 06:28 AM

Yeeah please send me some pics

would be graet

But if you remove STM filter, are you sure quality gets better, CQ value will get lower

//Wolfi

Wolfi · #**245** 05-01-2003, 06:29 AM

Quote:

Originally Posted by girv

Quote:

Originally Posted by Wolfi

I tried your script with the same movie

Kwag's script gave me a CQ value of 61.xx but your script was quite different, it gave me a CQ of 53.xx

Are you sure? Any source I've tried has come out smaller with my script at the same CQ...

Oh yeah I'm sure

//Wolfi

Kane · #**246** 05-01-2003, 06:50 AM

Quote:

Originally Posted by Wolfi

Yeeah please send me some pics

would be graet

But if you remove STM filter, are you sure quality gets better, CQ value will get lower

//Wolfi

well, i am sure, i´ve made many samples , and the sample with STM were always worse

Wolfi · #**247** 05-01-2003, 07:05 AM

Quote:

Originally Posted by Kane

Quote:

Originally Posted by Wolfi

Yeeah please send me some pics

would be graet

But if you remove STM filter, are you sure quality gets better, CQ value will get lower

//Wolfi

well, i am sure, i´ve made many samples , and the sample with STM were always worse

What CQ value did you use on your samples, if you reamember

around 60 maybe?

//Wolfi

Kane · #**248** 05-01-2003, 07:45 AM

i didn´t use a stable cq value

i let ToK do the prediction with both scripts, with and without STM and then i did a sample of the same movie ( i just made a 3minute sample with dvd2avi ) with the cq value the prdiction gave me, so i could compare the quality, the final movie would have

jorel · #**249** 05-01-2003, 07:54 AM

Quote:

Originally Posted by Kane

i didn´t use a stable cq value

i let ToK do the prediction with both scripts, with and without STM and then i did a sample of the same movie ( i just made a 3minute sample with dvd2avi ) with the cq value the prdiction gave me, so i could compare the quality, the final movie would have

right Kane,i did using the samples more or less like you.

please,remove my doubt:

in my (lots) samples,with STM is slow but better!
what is better for you?
with or without STM filter?

Kane · #**250** 05-01-2003, 08:00 AM

as i said before, the samples without were always better in my test.
as someone said before, it´s like looking through a big glass, the video is really unsharp with STM

Wolfi · #**251** 05-01-2003, 08:27 AM

Oke, if I drop STM filter, you think I still should have DCT filter

//Wolfi

Kane · #**252** 05-01-2003, 08:33 AM

Quote:

Originally Posted by Wolfi

Oke, if I drop STM filter, you think I still should have DCT filter

//Wolfi

i don´t use DCT
why always search for alternative filters , if you drop one?
sometimes less is more

Wolfi · #**253** 05-01-2003, 09:09 AM

Jorel, you think you can send some pics with STM and without, just to comapre, it would be great

//Wolfi

jorel · #**254** 05-01-2003, 12:22 PM

Quote:

Originally Posted by Wolfi

Jorel, you think you can send some pics with STM and without, just to comapre, it would be great

//Wolfi

of course my friend

send me a pm with your mail
and i send it later cos here in Brasil
on saturdays,sundays and holidays the internet is
tooo sloooooow.

....waiting your pm!

Wolfi · #**255** 05-01-2003, 03:51 PM

Thanks Jorel

//Wolfi

dazedconfused · #**256** 05-01-2003, 04:10 PM

@Kwag,

Could you please take a look at my post to andybno1 in this thread: http://www.kvcd.net/forum/viewtopic.php?t=3802&start=15 and tell me what you think of the script recommendations I listed there.

Personally, I don't think you should have dropped STMedianFilter completely from your Optimal Script. I know that a few people have complained that it reduces quality too much, but I happen to think it actually does a very good job at reducing filesize without effecting quality too much. Keywords: too much!

Because ANY extra filtering is going to cause some quality-loss probably. But for those longer movies where you really need some more compression, the first filter I would choose to add would actually be STMedianFilter. I did a few comparison tests (back at the beginning of this thread I think) between STMedianFilter and SpaceDust, and I found STMedianFilter actually looked a bit better. Just my opinion though.

Thanks,
-d&c

Jellygoose · #**257** 05-01-2003, 04:15 PM

Well I haven't made any tests in the couple of days, but I have had very good results with STM filter, just like jorel...
However we are using the max. Values for Spatial Filtering in STM so this might lead to an unsharp picture...
For me the question is just, if it's better to kill sharpness with a Spatial Filter or with a Luma Blur... ?? I still doubt that blurring the picture with a simple command will replace almost the whole temporal AND spatial filtering process...
This would somehow be too easy. I like to compromise for some reason...

kwag · #**258** 05-01-2003, 05:12 PM

I've done several tests today, and I came to the conclusion that the filter set like this: STMedianFilter(10, 30, 0, 0, 10, 30), I can't see any visual difference with or without the filter, but the file size is still smaller with the filter using these settings. Please try it

I've updated the current script again.

-kwag

Kane · #**259** 05-01-2003, 05:16 PM

ok, i´ll give it a last try, if it failes, i´ll delete the filter from my hdd

Wolfi · #**260** 05-01-2003, 05:19 PM

I will try this script on Showtime, 90 min

ConvertToYUY2()
LegalClip()
GripCrop( width=544, height=480, overscan=3 )
GripSize(resizer="BicubicResize")
STMedianFilter(10, 30, 0, 0, 10, 30)
SpaceDust()
unfilter(50,50)
temporalsmoother(1,2)
mergechroma(blur(1.5

)
mergeluma(blur(0.3))
DctFilter(1,1,1,1,1,.5,.5,0)
GripBorders()
LegalClip()

Just hope that ToK will work

//Wolfi