Using kvcd templates in different encoders

[trenchy]

I use mainly kvcd templates in tmpencoder to make vcd's out of divx.
However these conversion takes a lot of time compared with mainconcept encoder.
Is it possible to use the kvcd templates in other encoders like
mainconcept, cce or others?

Please respond if this is possible

trenchy

[kwag]

Hi trenchy,

You can use the "KVCD Parameters and Methods" with any encoder that has facilities to make changes to the Quantization Matrix, and allows for longer than standard GOPs.
You can use the matrix with MainConcept and CCE, and you will notice the increase in quality and file size reduction immediately. You can't use our parameteres with Canopus ProCoder, because it doesn't allow for changes on the matrix. You can also use the matrix with any hardware encoder that allows such changes, like the ones used for DVB transmissions over satellite. These are high end real-time encoders, usually at a distance from out pockets

-kwag

[trenchy]

Thanks for this fast reply kwag,

But i cannot just load the kvcd template in mainconcept. Or am i wrong
How should i use the kvcd template in mainconcept.
Do you have a guide for this?

trenchy

[kwag]

This thread should help you http://www.kvcd.net/forum/viewtopic.php?t=2891&highlight=mainconcept

-kwag

[BJ_M]

just look at the settings in the templates and copy them over to main concept -- you can input custom matrix tables and such (GOP structure) .. save as template ..

some of the svcd templates look interisting but I ran tests on a dvd templates and they produced non compliant streams for DVD (though will likely work in some players)

[txpharoah]

I've tried some of the kdvd/kvcd templates, and they're not compliant and not all that great if you're doing anything other than copying DVDs and trying to squeeze them on one disc. For your own captures, do something else.

[kwag]

I've tried some of the kdvd/kvcd templates, and they're not compliant and not all that great if you're doing anything other than copying DVDs and trying to squeeze them on one disc. For your own captures, do something else.

Really
Here's an old analog satellite capture, encoded with an old KVCD template: www.kvcd.net/samples/hbo-cap2.mpg
And the current results are way above that in quality 8)
And that's just peanuts! Take a look at what's currently cooking at kvcd.net: http://www.kvcd.net/forum/viewtopic.php?t=3999 BEWARE! very advanced topic

-kwag

[kwag]

I've tried some of the kdvd/kvcd templates

For KVCDs, there are workarounds to make them play on almost all players But yes, on some models they just don't work.
If you can't play a KDVD encoded MPEG-2 in any standalone DVD player, then you don't know what you're doing They are as compatible as the standard DVD MPEG-2's created with TMPEG's DVD templates.

-kwag

[trenchy]

Hi,

These link are mostly for KDVD do you also have ready made templates
for vcd in mainconcept encoder

trenchy

[BJ_M]

I've tried some of the kdvd/kvcd templates

For KVCDs, there are workarounds to make them play on almost all players :!: But yes, on some models they just don't work.
If you can't play a KDVD encoded MPEG-2 in any standalone DVD player, then you don't know what you're doing :!: They are as compatible as the standard DVD MPEG-2's created with TMPEG's DVD templates.

-kwag

kwag -- you have a lot of good information at your site and i do not dispute that -- but as one who not only creates content professionally but also helped develop those "standards" - I do know what I am talking about .. If you would like I can post the stream analys done from both hardware and software analysers ....

dont feel defensive though -- many of the software encoders bandied about here also fail in some areas -- and/or a lot of popular settings are not throughly understood/set correctly.

as i stated - most players can play (some can play stuff way out of spec) .. its just that I have to send out encoding 100% in spec (i admit in some areas the specs are rather loose) .. I do encoding not only for DVD but also for High Def and all types of d-cinema compression (including theatre/equipment design) (including color/image correction and compositing) as well as a consultant for a lot of film projects from 35mm to 15/70mm large screen ..

did I say not to use the information at your site ? NO ... in fact I helped someone who wanted to use a template in another program ..

so keep up the good work .. I follow whats going on ..

[kwag]

Hi BJ_M,

My comment was really to clear up the misunderstandings that I usually find around the forum, related to KDVD templates. You see, and you already know this but some people don't, is that the MPEG group defined the standards with some constraints and margins. Like the MAX size of GOP, etc. But what was never defined in the standard was a specific fixed Q. Matrix
This is exactly where we have the greatest advantage, and I emphasize that if any MPEG-2 encoder creates a DVD compliant file, and that file plays flawlessly on X DVD player, then the same encoder if encoding with our matrix, will also produce a 100% compliant MPEG-2 file that will also play on that X DVD player. If you analyze the MPEG-2 streams, you will see a difference in both encodes, but both will be within the MPEG-2 constraints, and should play in any DVD player. If a DVD player doesn't play a KDVD MPEG-2 stream, then by all means dump it in the trash can , because I guarantee that it will not play every commercial DVD on the market because of poor design. Like some of the first APEX models, I recall ALL kinds of problems with some DVDs. Some would play, some would lock up, and others simply didn't play at all

-kwag

[SaSi]

The MPEG-1 & -2 standards were defined in a refreshingly clever way: The standard defines how a decoder should decode a stream; rather than how an encoder should produce it. It is even stated that this was done to allow room for inovative approaches in encoding applications.

My view is that KWAG is "filling" this room for innovation by tweaking the parameters some encoders allow. In principle, only good can come from his approach.

At least for me, studying the way the Q.M coefficients are defined in the KWAG templates was enlightening in understanding what Quantization does to the frame.

On the topic: Quality vs bitrate vs speed, I have come to the conclusion that an enormous benefit can come in the MPEG encoding s/w by improvements in the Motion Vector determination algorithms. It appears to me that Tmpgenc has perhaps the slowest (perhaps brute force or full search) algorithms and perhaps CCE has the most clever in use. I have also noticed that a large number of (mainly Japanese) mathematicians are researching this very subject as we speak.

I would very much like to know both BJ_M's and KWAG's view on this. Is it true that motion search is the slowest part of the encoding process?

And, KWAG, Mainconcept allows a very wide range of parameters to be manually defined (via a "template"). It may be that Mainconcept's encoder is a more versatile platform for experimentation than Tmpgenc.

[kwag]

The MPEG-1 & -2 standards were defined in a refreshingly clever way: The standard defines how a decoder should decode a stream; rather than how an encoder should produce it. It is even stated that this was done to allow room for inovative approaches in encoding applications.

And that last sentence is exactly what describes KVCD

My view is that KWAG is "filling" this room for innovation by tweaking the parameters some encoders allow. In principle, only good can come from his approach.

Hopefully, we'll keep tweaking the hell out of the matrix, and any other adjustable facilities in the encoder portion of the MPEG protocol.

I would very much like to know both BJ_M's and KWAG's view on this. Is it true that motion search is the slowest part of the encoding process?

Motion estimation is one of the slowest parts for any encoder, and that's the key for qualty too. I agree that TMPEG has a slow algorithm, but for MPEG-1, it's probably the best encoder available on the market I've yet to see an encoder that can match TMPEG's quality. So I'll stick with it, even if it's slower. MainConcepts does a pretty good MPEG-1 job too, and it's a rocket in encoding speed. Wonder what algorithm they use, but it's fast! So there are many places for improvements

And, KWAG, Mainconcept allows a very wide range of parameters to be manually defined (via a "template"). It may be that Mainconcept's encoder is a more versatile platform for experimentation than Tmpgenc.

There's a thread at kvcd.net, related to MCE using our matrix. I haven't done that many tests with MCE yet.

-kwag

[txpharoah]

Sorry kwag, but open GOPs and several other aspects of your DVD encodes are NOT perfectly compliant. I also make DVDs professionally, and they MUST be complaint to ALL PLAYERS EVERYWHERE. Yours do not accomplish this. As I've said before, if somebody can get your template to work on the player THEY HAVE NOW without regards to future players or other people's players, then go for it. I'm all for you tweaking your encodes, but don't try to sell me a Chevy Nova and tell me it's a Chevy Corvette. They are great, not perfect. And many people like myself won't find a use for them. Not everybody in these forums just make things for themselves or to backup movies.

You don't have to be so defensive. Just learn to take a bit of criticism.

[kwag]

Sorry kwag, but open GOPs and several other aspects of your DVD encodes are NOT perfectly compliant. I also make DVDs professionally, and they MUST be complaint to ALL PLAYERS EVERYWHERE.

LOL, Have you ever opened a commercial VOB with bitrateviewer or other MPEG analyzer, and see how many closed GOPs it has ?
Closed GOPs are for easier editing of MPEG streams. They can be left open for final distribution, and THAT'S the way most (if not all!) commercial DVDs hit the street. As a matter of fact, some GOPs are open and some are closed, so most VOBs have mixed open/closed VOBs.

-kwag

[adam]

Well, the DVD standard certainly supports open GOPs. Closed GOPs are very inefficient when it comes to bitrate distribution, so they are only used on multi-angle video or when editing is needed, like Kwag said.

HOWEVER! I have noticed a big bug in TMPGenc where it doesn't maintain the set GOP length when GOPs are open. Obviously this shouldn't happen, but when it does it can in fact make the output non-compliant. So closed GOPs are not required for DVD compliancy per se, but when encoding in TMPGenc they almost might as well be.

[kwag]

@Adam,

I believe the bug you mention happens if you leave the MAX number of frames per GOP set to "0" (Automatic).
I have never experienced this, because I specifically set all templates to a MAX GOP size of 24, 25 or 30 (For KVCD templates, depending on the frame rate) and 15(PAL) or 18(NTSC) on the KDVD templates. This ensures that the number of frames will never exceed the MAX specified.

-kwag

[adam]

I've actually never used the Automatic GOP length setting. I've been able to reproduce this bug on numerous occasions. Simply loading the default NTSC DVD template and unchecking closed GOPs often gives me a non-compliant mpg. Scenarist gives me an error about having too many frames per GOP and all is well if I re-encode with closed GOPs checked instead. My guess is that the author of TMPGenc is aware of this bug otherwise I can't, for the life of me, conceive of why his default DVD templates would use closed GOPs.

[kwag]

Could it be the muxing
I never use TMPEG's internal muxing. I always use BBMpeg or Mplex.
I've encoded MPEG-2 and even MPEG-1 tricked with DVDPatcher to fool resolution, and I've burned the mpegs to DVD without any problems

-kwag

[adam]

Scenariest works with elementary streams not program streams. It does its own muxing, but not with TMPGenc encoded assets. It won't even accept them the majority of the time unless I close GOPs.