digitalFAQ.com Forums [Archives]

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)
-   Video Encoding and Conversion (http://www.digitalfaq.com/archives/encode/)
-   -   KVCD: Interlaced or not? (http://www.digitalfaq.com/archives/encode/5137-kvcd-interlaced.html)

holgerschlegel 08-22-2003 02:19 AM

KVCD: Interlaced or not?
 
Hi,

I've captured a movie via tv-capture card at 480x576 to an avi file. If I open the avi file as source in tmpg the source video type is set to interlaced. Because of the resolution both fields are captured and so interlaced is correct.

If I wrap the avi file into an avs script containing a call to FieldDeinterlace right after the AVISource line, the output of the avs should be an deinterlaced movie, right? But why is the source video type set to interlaced after loading the avs into tmpg?

Holger

Dialhot 08-22-2003 04:01 AM

I'm not sure TMPGENC can detect automatically what is interlaced and what is not.

kwag 08-22-2003 10:59 AM

Re: interlaced or not
 
Quote:

Originally Posted by holgerschlegel

If I wrap the avi file into an avs script containing a call to FieldDeinterlace right after the AVISource line, the output of the avs should be an deinterlaced movie, right?

Yes.
Quote:

But why is the source video type set to interlaced after loading the avs into tmpg?
You probably loaded a template "after" you had loaded your .avs file, so you are seeing the default template settings.

-kwag

incredible 08-22-2003 11:39 AM

Re: interlaced or not
 
Quote:

Originally Posted by holgerschlegel
Hi,

I've captured a movie via tv-capture card at 480x576 to an avi file......If I wrap the avi file into an avs script containing a call to FieldDeinterlace right after the AVISource line, the output of the avs should be an deinterlaced movie, right?
Holger

Why do you deinterlace if you encode your avi to mpg at still 576 high size?
Well on the one hand you can deinterlace if your Tv-source is good, like Tv-Commercials etc. But if you capture a blockbuster like from PRO7 (german tvStation) you often receive more grainy and unsharper captures. Compare, you will see! ;-) And than you can make this:
In the deinterlacing process you will obtain quality loss, even by using a adaptive deinterlacer. Try to encode your Video using mpg2+interlaced @ 480x576 with no deinterlacing in the AVS Script and do your prediction.
BUT caution if you determine your video size using Moviestacker! If you choose overscanning than at least use overlap-overscanning! By using resize-overscanning the orig interlaced Fieldorder will be destroyed cause of also vertical resizing afterwards and everything was for nothing ;-)

kwag 08-22-2003 12:27 PM

Re: interlaced or not
 
Quote:

Originally Posted by incredible
BUT caution if you determine your video size using Moviestacker! If you choose overscanning than at least use overlap-overscanning! By using resize-overscanning the orig interlaced Fieldorder will be destroyed cause of also vertical resizing afterwards and everything was for nothing ;-)

Hi incredible,

Can you verify this :?:
Because to my understanding, the image will always be resized on the TV, to do the overscan blocks. So the image is always streched both horizontally and vertically. So if we encode with "Resize" or "Overlap", the final image is still going to be stretched horizontally AND vertically on the TV, no matter what, because it's an internal TV function. :?: :?

-kwag

incredible 08-22-2003 02:13 PM

Re: interlaced or not
 
Quote:

Originally Posted by kwag
Hi incredible,

Can you verify this :?:
Because to my understanding, the image will always be resized on the TV, to do the overscan blocks. So the image is always streched both horizontally and vertically. So if we encode with "Resize" or "Overlap", the final image is still going to be stretched horizontally AND vertically on the TV, no matter what, because it's an internal TV function. :?: :?

-kwag

Well lets see if I can explain in english...

The basic (for example PAL) interlacing is build on 50 Lines at 50Hz which give you 25 FPS (pictures) PAL for example @ 704x576 and thats the point. In (full PAL) digitally captured pictures you got an interlaced image at 576 high if you scale down (not cropping) the size for the amount of 50 Lines proportionally to the space is trashed. (Its like dots per inch in Photoshop, .. I only use this comparison to explain ;-))
You can let out every 2nd line to obtain 288 px but also then its not interlaced anymore but also correct handled cause you did not "resize" to 288 but you let out every 2nd line and thats for example they way a direct capture at 352x288 works (PAL). If you capture at 768x576 and want to scale to for example 640x480 you HAVE to deinterlace before.
What we want to receive is that the architecture of the 50 Lines at 576px of picture high will be untouched and we can if we only resize the widh to 480 .. so we get a Picture at 480x576 which is still interlaced and in this untouched state it appears more sharpen than an interlaced one.
Ok by watching it on a PC you can see the comb effekt, but who cares, cause our intention is to make CDs watchable with a DVD Player ... and not CD with Dvix which will only be seen on a PC Screen.
But you have to use for example asharp with caution! Cause asharp sharpens horizontally and vertically... and if you sharp an interlaced picture at is vertcal to much .. you can see flickering when watching on Tv!
So it would be better to make a combination of asharp() and unfilter(only horizontally) .. this will give good results.
The picture then will not be scaled in the DVD Player internal as we "digitally" think, no ... it will be converted using the d/a converters to give an analog signal to the Tv and the Tv scales the "analog" picture as a whole and thats the difference. So every Tv scales the picture more or less and thats why they use the "cache" which normaly takes about 10% at the borders of the screen .. calles overscanning ;-)
So we put on the untouched (at the high *576*) stream a black layer at the borders .. and this we call overlapping.
So with all this we get a correct interlaced 3/4 PAL 480x576 interlaced picture including correct set overscanning borders, ready to convert to mpeg2 which can handle interlaced frame architecture.

Maybe someone sees now a misunderstanding, but I tried to explain this complex materie in my best english words ;-)

holgerschlegel 08-24-2003 04:25 AM

Re: interlaced or not
 
Quote:

Originally Posted by kwag
Quote:

Originally Posted by holgerschlegel

If I wrap the avi file into an avs script containing a call to FieldDeinterlace right after the AVISource line, the output of the avs should be an deinterlaced movie, right?

Yes.
Quote:

But why is the source video type set to interlaced after loading the avs into tmpg?
You probably loaded a template "after" you had loaded your .avs file, so you are seeing the default template settings.

-kwag

No, I've first loaded the template and then open the source avi file.

The reason why I deinterlace is, that I want to encode to mpeg-1. Read more then once that mpeg-1 is better on low bitrates which I must to use to fit 90-120 minutes to once 80min cdr.

Holger

incredible 08-24-2003 07:01 AM

Re: interlaced or not
 
Quote:

Originally Posted by holgerschlegel
The reason why I deinterlace is, that I want to encode to mpeg-1. Read more then once that mpeg-1 is better on low bitrates which I must to use to fit 90-120 minutes to once 80min cdr.

Holger

You're exactly right to receive better compressing results by using mpeg1 ... the way above is a compromise to encode a movie like 90 min with best quality in sharpness etc. .. It depends upon the CQ you reach by doing a prediction before. If you get a good CQ you can encode using mpg2 in interlaced mode, otherwise ... deinterlace (I've put a capture script in the Thread Script for captures and vhs in the scripts-topic) and encode using mpg1.

The CQ you reach is not only based upon the mpeg encoding method you choose. Also it depends upon the "real" movie information within the whole picture (fullscreen, widescreen, cinemascope) the more borders you have, .. the more you will receive a higher CQ. And ... it depends also upon what "happens" during the movie.
Example:
Matrix= ca. 120min / much dark scenes/ 80min-CDR used = ca. CQ 62
Message in a Bottle: = ca. 115 min / very much changes in light and dark = ca. CQ 46!!

Both predictions came with the same MA Script used! Same conditions same mpeg etc ...


All times are GMT -5. The time now is 11:48 PM  —  vBulletin © Jelsoft Enterprises Ltd

Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.