DVD2SVCD: AVI Conversion-> Fixed Subtitles are "Cut Off" ?

[Paulus]

Sorry....this probably is a stupid question......

I have a .avi file that has fixed subs in it which are partially in the image and partially (second row) in the lower black border. Avi is 640x480.....

When I use the DIVX4 script from the D2S optimal script section the subs are cut off horizontally (half of the upper line of the subs is cut off and the lower sub line is completely gone in the endresult .mpg file.).

Also during encoding this is visible in TMPG....

Can I adjust the script somewhere to test tell D2S to encode a bigger section of the source .avi (Gripborders ?).

[Dialhot]

Quote:

Originally Posted by Paulus

Can I adjust the script somewhere to test tell D2S to encode a bigger section of the source .avi (Gripborders ?).

That is Gripfit that cut of the black bars without noticing the subtitles. You must modify the script in ordoer to use the normal resizing that is done by D2s (that means, remove the gripcrop/gripsize/griborders lines and put at the same places the code you can copy from the bicubic script that is provided by D2S.
Don't forget to number all the lines correctly at the end.

[Paulus]

Thnx Phil,

Will try it tonite............

[Paulus]

Need a little extra advice......

I'm trying the script below as we speak and the picture in TMPG seems good. Can I put an "overscan=1" line somewhere maybe to get lower filesize ?

Additional question: ........my source avi seems to have a little bit too dark picture. Can I increase "brightness" with some script line ?


[AVISYNTH_Divx FilterV4->Avi's+PermSubs]
0=BlindPP(cpu=4)
1=Blockbuster(method="noise",detail_min=1,detail_m ax=3,variance=0.1,seed=1)
2=Convolution3D(1, 6, 12, 6, 8, 2.8, 0)
3=BicubicResize(^TargetWidth,^TargetHeight,^b_valu e,^c_value)
4=Undot()
5=TemporalSoften(2,7,7,3,2)
6=DCTFilter(1,1,1,1,1,1,0.5,0)
7=function ConvertToYUY2(clip c){ return isYV12(c) ? c : convertToYV12(c) }
^b_value=0.0
^c_value=0.6

[Dialhot]

Quote:

Originally Posted by Paulus

Need a little extra advice......

I'll give you the best one : STOP USING D2S for making avi.

That's a complete nonsense are we all know how D2S is inconsistent in file size prediction. There is no benefit's to use it where notepad, Moviestacker and CQMatic is enought.

It's time to move over and use CQMatic !

Use D2S only for DVD.

Quote:

I'm trying the script below as we speak and the picture in TMPG seems good. Can I put an "overscan=1" line somewhere maybe to get lower filesize ?

You can't. Second reason to stop using D2S.

Quote:

Additional question: ........my source avi seems to have a little bit too dark picture. Can I increase "brightness" with some script line ?

Add a "Levels(0, 1, 255, 0, 255)" in the last line. The more you change the "1", the more you increase brigthness without making the picture too "white" (1 is the gama value. 1 = neutral, go by little step, 1.2 should be enought).

Quote:

[AVISYNTH_Divx FilterV4->Avi's+PermSubs]
0=BlindPP(cpu=4)
1=Blockbuster(method="noise",detail_min=1,detail_m ax=3,variance=0.1,seed=1)
2=Convolution3D(1, 6, 12, 6, 8, 2.8, 0)
3=BicubicResize(^TargetWidth,^TargetHeight,^b_valu e,^c_value)
4=Undot()
5=TemporalSoften(2,7,7,3,2)
6=DCTFilter(1,1,1,1,1,1,0.5,0)
7=function ConvertToYUY2(clip c){ return isYV12(c) ? c : convertToYV12(c) }
^b_value=0.0
^c_value=0.6

??? You forgot the addborders line ! (put it in 8 and the "level" command in 7, then the function line in 9).

[Paulus]

Hey Phil,

Not so "negative" on D2S my friend .....

Filesize prediction is not a big issue for me. In 2 or three encodes I usually end up with a 800 Mb .bin file. No problem.

That "overscan" isn't possible is a pitty. Don't know how much it affects the final filesize, though....

But I WILL take your advise into consideration and start the manual procedure with next avi project.......(expect more questions from me.. ).

About this (last) D2S Avi conversion........

I did have Addborders before (I think "AddBorders(0,^BorderTop,0,^BorderBottom)....b ut that resulted in a "squashed" picture. Do I need to add something like "sourceana=false" ?


Paulus......

[Dialhot]

Quote:

Originally Posted by Paulus

I did have Addborders before (I think "AddBorders(0,^BorderTop,0,^BorderBottom)....b ut that resulted in a "squashed" picture. Do I need to add something like "sourceana=false" ?

No, D2S do that itself when it computes BorderTop and BorderBottom. In fact it takes every avi as "non anamorphic".
The picture was squashed in the D2S preview ? If yes, that's normal (this preview does not deliver a a/r corrected picture).

[Paulus]

Hi Phil,

Picture appears to be squashed in Tmpg window. So I think something is wrong......

I know D2S preview screen always gives a wrong imprission, I wasn't talking about that.

What should be changed in the script ?

(now is :

0=BlindPP(cpu=4)
1=Blockbuster(method="noise",detail_min=1,detail_m ax=3,variance=0.1,seed=1)
2=Convolution3D(1, 6, 12, 6, 8, 2.8, 0)
3=BicubicResize(^TargetWidth,^TargetHeight,^b_valu e,^c_value)
4=Undot()
5=TemporalSoften(2,7,7,3,2)
6=DCTFilter(1,1,1,1,1,1,0.5,0)
7=AddBorders(0,^BorderTop,0,^BorderBottom)
8=function ConvertToYUY2(clip c){ return isYV12(c) ? c : convertToYV12(c) }
^b_value=0.0
^c_value=0.6

[Dialhot]

Quote:

Originally Posted by Paulus

What should be changed in the script ?

The script seems okay to me. Check in D2S that you had selected "4:3" and not "16:9 encoded as 4:3 (borders added)"

[Paulus]

I always use "4:3 (No borders, encode as 4:3)" for KVCD......I learned my lessons with that

Still the picture showing in TMPGenc looks wrong / squashed.

Maybe I have to remove my "AddBorders" line because the avi already has borders in it (with subs in the lower black bar). Without the AddBorders Line the picture in TMPGenc looks OK but........at the bottom of the TMPGenc screen it shows "DVD2SVCD (MPEG-1 352x384 25fps CQ 84)" That is a strange resolution "352x384" because I chose CVD resolution 352x576..........hmmm......confused......

[Dialhot]

Quote:

Originally Posted by Paulus

at the bottom of the TMPGenc screen it shows "DVD2SVCD (MPEG-1 352x384 25fps CQ 84)" That is a strange resolution "352x384" because I chose CVD resolution 352x576..........hmmm......confused......

You have these because you missed the borders (that are (576-384)/2 pixel each).

What is the resolution of the source ?

[Paulus]

source = 640 x 480 (see first posting)......

Each border is 576-384=192 / 2 = 96............

[Dialhot]

Quote:

Originally Posted by Paulus

source = 640 x 480 (see first posting)......

*
Perfect 4:3 source. I do not see there what make D2S fails.

2 issues : either the picture isn't distorted as you think it is, or D2S is a tool to drop.

Make your choice...

[Paulus]

Phil,

It appears to be the CVD resolutions I'm using is causing the problems.....

If I use SVCD res (480x576) the picture that appears in TMPGenc screen is OK.

....I've read somewhere that bicubic resize can cause troubles with CVD res....(?).

Will post again if project is finished....

(and than start using CQ matic...........)