Quantcast KVCD Template Confusion - digitalFAQ.com Forums [Archives]
  #1  
12-10-2003, 07:45 AM
radunn radunn is offline
Free Member
 
Join Date: Nov 2003
Posts: 26
Thanks: 0
Thanked 0 Times in 0 Posts
I'm capturing vhs in Virtualdub-VCR at 512*384, encoding (frame-serving) with KVCD template (NTSC-film_plus)

1. Am I correct to say that Deinterlacing should not be done during the capture; only during the encoding process? And, is it true that deinterlacing is only necessary when resizing down from the capture (the vertical value) and not needed if encoding at "XXX x 240"?

2. When I loaded the captured AVI directly into TMPGEnc it stopped recording video/audio at about 25%. When I frame-served from Virtualdub it was fine; what is the issue here.

3. (and last) is that using KVCD template (NTSC-film_plus) Project Wizard (TMPGEnc) by default enabled Inverse Telecine; should this be unchecked. I removed the checkmark and the fps was encoded at 29.97fps (not the default template value).

This is a dynamite tool; one that has gotten my interests.
__________________
Radunn
Reply With Quote
Someday, 12:01 PM
admin's Avatar
Site Staff / Ad Manager
 
Join Date: Dec 2002
Posts: 42
Thanks: ∞
Thanked 42 Times in 42 Posts
  #2  
12-10-2003, 09:37 AM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
First of all, you should capture at 512x480 to capture both fields.

You'll have to deinterlace in every case if you encode to MPEG-1. Deinterlacing during capture is a no-no, you'll get better quality if you deinterlace during postprocessing.

Use Avisynth for frameserving to get the max out of KVCDs.
Reply With Quote
  #3  
12-10-2003, 09:50 AM
radunn radunn is offline
Free Member
 
Join Date: Nov 2003
Posts: 26
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
First of all, you should capture at 512x480 to capture both fields.

Use Avisynth for frameserving to get the max out of KVCDs.
To be sure, Deinterlacing in TMPGEnc is at Deinterlace (none)?

What was/is the result of capturing at 512x384 (it 'was' recommended if disk space is an issue etc.).. but I'm here to learn what IS right.

I am not that adept on using Avisynth (scripts); doesn't Virtualdub handle frameserving well? Thanks for yojr support
__________________
Radunn
Reply With Quote
  #4  
12-10-2003, 10:21 AM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
I don't know exactly what happens if you don't capture at n x 240 or n x 480, but I know that you'll definitely lose valuable information even when capturing from VHS sources. If diskspace is the issue, you should rather capture at 480x480 (or even lower horizontal res) than 512x384 - the vertical resolution of 480 pixels is the most important when capturing.

Avisynth is faster than VDub and you'll also get to use better filters. For scripts check out the Optimal scripts forum, there's a huge amount of information over there.

If you deinterlace in Avisynth/VDub, you have the correct setting in TMPGEnc.
Reply With Quote
  #5  
12-10-2003, 02:56 PM
radunn radunn is offline
Free Member
 
Join Date: Nov 2003
Posts: 26
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by Boulder
If you deinterlace in Avisynth/VDub, you have the correct setting in TMPGEnc.
I thought you said that it's a no-no to deinterlace before the final encode; or are you referring to frame-serving?
__________________
Radunn
Reply With Quote
  #6  
12-10-2003, 03:00 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by radunn
or are you referring to frame-serving?
I'm pretty sure that's what he meant

-kwag
Reply With Quote
  #7  
12-10-2003, 03:57 PM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by kwag
Quote:
Originally Posted by radunn
or are you referring to frame-serving?
I'm pretty sure that's what he meant

-kwag
Yep, postprocessing TV captures is basically deinterlacing(unless encoding as interlaced for MPEG-2 use)+cropping+filtering+resizing in Avisynth and then feeding to TMPGEnc/CCE.

I wouldn't let TMPGEnc do anything apart from encoding the video
Reply With Quote
  #8  
12-11-2003, 08:09 AM
radunn radunn is offline
Free Member
 
Join Date: Nov 2003
Posts: 26
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by kwag
Quote:
Originally Posted by radunn
or are you referring to frame-serving
-kwag
Yep, postprocessing TV captures is basically deinterlacing(unless encoding as interlaced for MPEG-2 use)+cropping+filtering+resizing in Avisynth and then feeding to TMPGEnc/CCE.
Hmmm "unless encoding as interlaced for MPEG-2 use;" do you mean as DVD/SVCD? MPEG-2 format should "not" be deinterlaced? I've never worked with DVD yet; only SVCD. Again, thanks!
__________________
Radunn
Reply With Quote
  #9  
12-11-2003, 09:02 AM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
He means that IF you encode using MPEG2 (which supports interlaced encoding) you should leave it interlaced cause of preserving more quality.
But on the other hand Mpeg2 interlaced needs higher bitrates (IMHO).

@Kwag/Boulder

... we maybe should try to convert the NOTCH matrix to an alternate scan order?? What's you point of view on this?
As many people keep their interlaced material to encode in KVCD Notch Matrix mode
Reply With Quote
  #10  
12-11-2003, 09:19 AM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
Quote:
Originally Posted by incredible
He means that IF you encode using MPEG2 (which supports interlaced encoding) you should leave it interlaced cause of preserving more quality.
That's correct. Motion is smoother and the image sharper this way. Just make sure you've got the correct field order

Quote:
But on the other hand Mpeg2 interlaced needs higher bitrates (IMHO).
Yep, the filesize increase can be quite large so more bits are needed.


Quote:
@Kwag/Boulder

... we maybe should try to convert the NOTCH matrix to an alternate scan order?? What's you point of view on this?
As many people keep their interlaced material to encode in KVCD Notch Matrix mode
This is something I'll gladly leave to Kwag I don't know much about the matrices, how they work etc. Maybe there is a simple way to alter the matrix to get one for interlaced material?
Reply With Quote
  #11  
12-11-2003, 12:43 PM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
@ Boulder

Every block of the 8x8 field (one block = 8x8 pixels) on where the DCT will be processed:


Here you can see the 8x8 scan order of an

ZIG-ZAG Order (farme based) and ALTERNATE Order (field based):


"MB1" did a matrix conversion to an Alternate scan
(I don't know which was the origin where this does base on, maybe the standard CCE Matrix as this is a screenshot of CCE) --- here is it:


This all bases on a very well explanation of "Kika" from Doom9.de/Gleitz, here's the source site, ready translated by using Googles Language Tools:
http://translate.google.com/translat...language_tools

Its very interesting and discovers a lot of how matrixes and DCT works.

But anyhow as we are here at KVCD to fit one movie on one CD-R I think the conclusion in THIS case could be that its better to deinterlace so first you can encode using mpeg1 and second you won't need that more needed Bits and therefore we can enjoy a lower AVG Bitrate.
And in case of preserving the max quality we should leave the interlaced state.
Reply With Quote
  #12  
12-11-2003, 03:09 PM
Boulder Boulder is offline
Free Member
 
Join Date: Sep 2002
Location: Lahti, Finland
Posts: 1,652
Thanks: 0
Thanked 0 Times in 0 Posts
Thanks inc, now I understand why the values in the matrices have a certain pattern - it's due to the scanning order. Which brings up a small idea : couldn't the KVCD notch matrix be altered to suit interlaced sources just by changing the values according to the graphs you attached?

I agree, deinterlace for KVCDs, encode as interlaced if you've got the bitrate to spare (such as KDVDs).
Reply With Quote
  #13  
12-11-2003, 06:13 PM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
YEP Boulder thats the way, but I never tried it cause of forced non-interlaced mpeg1 encodings in case of OneCD Encodings and in case of HighQuality Productions, like preparing a commercial advertising for business etc I always used MB1's alternate scan Matrix which on the other hand just filters a little bit the frequencies in comparison to the NOTCH.

My idea above to do an alternate scan version of the notch was at least to respect the scan order when the matrix does it's job on interlaced sources IF someone really wants to encode interlaced movies as a whole to one CD-R
Reply With Quote
  #14  
03-06-2005, 05:05 PM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Just to update this (very old! ) thread, the matrix (any matrix) is not dependant on scan order.
Scan order is determined by the encoder, but each matrix position (weight) works on a particular frequency range.
So it doesn't matter what scan order is used. The scan order is used depending if the source is interlaced or progressive, and this has nothing to do with matrix order.
I'm updating this thread because I still see some confusion on other web sites, which point to this thread.

-kwag
Reply With Quote
  #15  
03-07-2005, 05:45 AM
incredible incredible is offline
Free Member
 
Join Date: May 2003
Location: Germany
Posts: 3,189
Thanks: 0
Thanked 0 Times in 0 Posts
Send a message via ICQ to incredible
Karl, if you force the encoder to treat interlaced material correct by setting the scanorder to "alternate" then the encoder gots a different image to encode! Cause then the encoder does its DCT procedure on a fildbased, means fieldseparated Stream. Do make your own preview in Avisynth by using the command separatefields() and you see what the encoder gets in its alternate DCT routine. THATs why an interlaced/alterante scanorder matrix DOES make sense!

Reply With Quote
  #16  
03-07-2005, 06:16 AM
kwag kwag is offline
Free Member
 
Join Date: Apr 2002
Location: Puerto Rico, USA
Posts: 13,537
Thanks: 0
Thanked 0 Times in 0 Posts
Hi Inc,

Yes, I agree that if you force the encoder to use a different scan order for an incorrect source, the results will not be correct.
But the point is that each of the 64 blocks of the DCT matrix's frequency domain is fixed, no matter what scan order is used
That's what I'm trying to say
It doesn't matter if you scan left, right, up or down, that "position" of the DCT block will ALWAYS be applied to a specific frequency range.

-kwag
Reply With Quote
Reply




Similar Threads
Thread Thread Starter Forum Replies Last Post
Something to clear up confusion and misleadings. kwag Off-topic Lounge 13 11-02-2005 06:59 AM
TMPGEnc: Aspect ratio confusion Glebix Video Encoding and Conversion 1 12-31-2004 09:22 PM
Bitrates: Calcumatic average bitrate figure confusion?? Blubear Video Encoding and Conversion 11 11-13-2004 09:42 PM
KVCD: GOP numbers confusion? AdamJ Video Encoding and Conversion 11 08-19-2004 12:47 PM
confusion with TMPGEnc settings nicksteel Video Encoding and Conversion 24 12-18-2003 12:54 PM

Thread Tools



 
All times are GMT -5. The time now is 06:07 AM  —  vBulletin © Jelsoft Enterprises Ltd