Ok, I want to bring it up again, and hopefully after that we can burry it. I've been testing this baby too now, for several weeks. I also tested this setting in the "quantization type" setting...

Ok, I believe everbody now caught up with the fact that the stream produced by ffvfw is NOT an H.263 stream. What some have not seemed to realize yet is:

The stream produced by ffvfw with setting set to H.263 is the EXACT same as if you set it to MPEG2. That is what kwag insisted and that is what my tests confirm. Same size, same GOP structure, same Q-Curve, same bitrate allocation.

Over and out.

In BV it is 14 so realy 28.
I now try to figure out where it is VBV value seted in file. I compare two files with differend VBV under HexEdit but they are identical. It is mess of @!$@!&$%@$ like signs along with jahsdkjAHkjHASOIAS and some numbers and some ...... as usual in HexEdit :lol: so I migh overlooked this. I'm keep trying.


EDIT: Hold youre horses. I think I foud entry. Few more test. I'll be bec in 5-10 min.

Great stuff Hydeus.
I was hoping that the vbv had been corrected by 40 but in fact it was by 44 since now you have 72 minus the 28 you had previously equals 44.
I wonder if there is a way to tweak tmpg in the registry so that it patches the file to 224 instead of 44...
I'll take a look at it now, though I'm not a regedit expert...
Cheers

:lol:
Oh my god... you're trying to change that VBV Buffer size of a file with a Hex-Editor? :lol:
That is the funniest thing I've heard in a long time. I think the last time I used a Hex-Editor was to activate cheats in Doom II... :mrgreen:
No offense, seriously buddy... :wink:

Read my last edit :!:

@Hydeus,
You mean in the registry or using the hex editor 8O
Be a sports and be back in 1m :lol:
Cheers

BTW, I think that it might be of some use if we could establish, at some point, just for testing purposes, an unique set of settings (or maybe divided in two or four for VCD/KDVD/constant-bitrate/multiple pass) for an encoding and everybody interested make a movie with that settings and post the exact results/issues/failures with that encoding on his own player. I think that way we might have a more exact picture of the issues with this encoder regarding particular brands and models of dvd-players.
Since in making of a KVCD/KDVD we have to break some standards anyway, at least let's find out how far we can go. Moreover we could picture more exactly where the problem with ffvfw resides or could be done in the futhure. For example untill yesterday I lived under the impression that everybody could play at least a VCD made with my settings and I would have been under that impression even now if rds wouldn't point out his problems playing such a file on his player and others.

I just learned that Milan is back and I think it might be usefull for him such a poll. Besides, I'm not sure how much interest he has in mpeg2 encoding since a big deal of ffvfw is about mpeg4 encoding so I think that a suplimentary motivation for him would be good.
marcellus

Geez this is crazy... :lol:
What do you achieve by finding the Hex Values that set the VBV value :?:
It's not like it is at the same spot with every file you encode...

Realy :?: Files: one with VBV 40 and one with VBV 112 are differend in only two entries. ONLY TWO

Ive changed now succesfulu VBV from VCD to SVCD 8)
I'll post some instruction in few minutes.

@Hydeus: I totally see your point but...

The way I see it, in order to see the differences between 2 files refering to the VBV buffer you'll always need :

1 File with the wrong VBV Buffer +
1 File with the right VBV Buffer

The problem is, we can't produce a stream with the right VBV buffer with ffvfw right? So we can't know where the correction has to be done. I really do not believe that the information of the VBV buffer is at the same spot in generally all MPEG Files... Or is it? :roll:

Hi guys,
Couldn't agree more with you, Marcellus.
Do you want to tell us your latest settings?
Hey Hydeus? What about your's? Anybody else wants in with us?
While both you two post it your settings remember to tell us which build you're using so we can all tune up.
What do you think :)

Ok here we go :D

- Open file in hexeditor of any kind you like.
- first line is our goal. its a bit tricki so read carefuly. Migh be some errors, I'm nevbe as you all know ;)
it is build of 2 column , each of 8 pairs (hexadecimal)
oure goal are two pairs in second column afret FF FF
:arrow: For VBV 40 (20BV) type E0 A4
:arrow: For VBV 112 (56BV) type E1 C4
I don't preformed DVD VBV , so for now i dont know what value it will be , but i think it can be calculated.

Test this while it is still not all.

I'll be back in few minutes (of typing ; )

EDIT: Jelly we can produce files with proper VBV bufer, at least Kwag say it is possible. While I'm not expert on this I can only belive hes words.

Worst than that. What if the vbv is not only in the header of the file (where I think Hydeus caught it) but the whole way in which the "data" is registered is different.
That will give us a correct vbv under BV but a choppy playback because the stream is still vbv 14 in terms of body...
Hope you can prove me wrong Hydeus.
C ya

I can't cause I don't know anything more about this. But I've compared two files prduced by ffvfw. One has VBV of VCD and one of SVCD, and both are produced with this values by ffvfw, with no my modyfications. I only modyficate after encding. In fact with compare two files i've discovered where VBV value sets.

And thiss is where my problem lies. Same set of hexadecimal pairs repeats one more time in the stream. But now this is more complicated, while I dont know where to find them in some random/differend stream :cry: I'm still testing.

Maybe this is a dumb question, but why is there a vbv setting for muxing in avi2mpeg when this info is already present in the video stream?

I don't know the purpose of that, while video stream with VBV40 muxed as SVCD with 230 (another strange value ;) ), and demuxed still have VBV40.

Back to hexedit.
This is not one more time (I have 10s sample), I think this is period of GOP lenght. I need a specialist. I'm only newbie :lol:

A VBV buffer size entry, is present on every "Sequence Header" of an MPEG-2 file.
This occuurs about every half second in the stream.
So if anyone wants to patch a VBV buffer size, have fun :roll:

References: Video Demystified. Page 573.

Edit: I can see the writing on the wall: "kwag, can you write a VBVPatcher program" :lol:

-kwag

@rds_correia:
My settings I use since at least two weeks with good results are in this thread:
http://www.kvcd.net/forum/viewtopic.php?t=9191&start=0
I hope that everybody could see the pictures, I had an issue on that side but I think I solved yesterday. But anyway I think that a new thread is necessary for this test.
EDIT: I forgot to say that I use the latest version you can find at this address:
http://www.ligh.de/software/mirrors.phtml

@jellygoose
I don't really understand the point in discouraging or making fun of somebody because he has an idea that could look dumb for the moment but in fact could be very useful at some point. This is a basic rule in brainstorming, everybody can say anything, no matter how stupid might look and nobody can criticize it. In the end we'll eventually see if it was a good idea or not.
marcellus

Hya,
Go for it Kwag :lol:
And if it's not asking too much could you build it with bitrate constraint routines :?: :twisted:
Cheers buddy

So you say the same as me (AFI understand). At begin of GOP :?: So, if it is possible to know where GOP start (AFAIR ReStreem knows ;) ), aplication that changes VBV value would be possible to create. Where are you vHelp :twisted:


EDIT: And I see now that this is possible.
[Crazy enought programer wanted] :lol:

I even think that the GOP header is (more or less) 2416 bytes long.

Hi Marcellus,
I still can't see your screenshots.
I only see tripod's logo.
Damn thing it uses the registry instead of ini file or something.
Can you check your settings at tripod again?
Thanks

Don't push it :twisted:
Screen interface already done ;)

http://www.digitalfaq.com/archives/i...2004/02/36.jpg

Now I'll go read more on the MPEG-2 stream, so I can search every sequence header for the VBV offset, and patch it.
I hope this works :roll:
If not, it was worth a try ;)
DON'T EXPECT THIS OVERNIGHT, AND NO PROMISES :!: :rotf:
(I hope Milan shows up, and fixes it, instead of having to make this hack :roll: )

-kwag

Yep,
That's definitly vhelp :rotf:

Somebody call me :mrgreen:

Another brainstorm, and we all have that big impossible to resolve thing resolved by aplication egzisted from years (ok seconds ;) ) :D Do we have to make everything this way :evil:

http://www.digitalfaq.com/archives/i...2004/02/37.jpg

:rotf:

I want my credits in this :!: :hammer: :bugeyes:

@Kwag
Have you contacted Milan:?: I have no response (send 22:30 today).

You probably knows header lenght, but if not (I dont know either) your "red_planet_extreme.m1v" (I think it is still broadcasted by u) have very specyfic header consist of some ....R...JR".)H...". period, and not only this. Very beautiful, you must se this :lol:


I now go to sleep. God night to all, exept Kwag, you must finish this task :crashed: :expert: and after that you can :playball: a while :rotf:

No response here either :!: :lol:

rds_correia, your awesome!!
Thanks for smoothing things over....

I was just thinking about the very best program/movie to do tests on --- I can think of nothing better than the first episode of Blue Planet - Has anyone seen this program???

It has wealth of great scenes to test: 1) highly complex moving fish to test motion estimation, 2) underwater gradients, 3) slow moving pans of large whales (even slow motion photography). In my opinion, this puppy has everything...

Sorry guys...

I've ben very busy w/ my GUI app :roll: and haven't had the time to read
up on all the threads - just random checkings.

It's late and my :roll: are beginning to pop out.
-vhelp

Slow down marcellus. I did not discourage anybody. I think you're again the only one to feel offended by this.
Hydeus did not take that personally as you did. :wink:

I didn't take this personaly , cause I knew that I was right. And when Kwag will build this program (as far as he will buid it) it will be ...

IN YOURE FACE J3LLY PAL :lol: 8)

8)
And that's the way it should be. Sometimes you're right, sometimes you're not. Good luck kwag, and thanks for that idea hydeus... :mrgreen:

Well, I didn't take anything personal either :D
Hope the same about you :wink:
Just a thought that passed through my mind, nothing more...
8)
marcellus

:D That's the spirit guys :D
@Marcellus,
I don't know if I'll have enough time before the weekend but I really would like to try your ffvfw settings.
Any luck on the hosting yet?
Cheers

As promised, here's version 0.1 ;)

http://www.kvcd.net/VBVPatcher-0.1.exe

Notes:

It only patches the FIRST sequence header, so the rest of the file is still unpatched.
PLEASE, i've tried to do my best, and force to open only a .m2v file, and even if you rename a file that is not a Video Elementary Stream to a .m2v, I check the header and display that it's not a .m2v file.
This is to offer protection, and you don't screw up a file that is not a .m2v video stream :cool:

Please test it, by creating some mpegs with ffvfw, and then patching it.
After you drag the mpeg file to Bitrate Viewer, you'll see thah the VBV is 112.

Next version will enable to select different VBV values, and will patch the complete file.

But first, let's play with this for a little bit, ok! :)

-kwag

I begin to be embarassed by this stuff. When I open the thread I have no problem seeing the pictures. How is this possible, that I can see them and you don't? I am a newbie in the free hosting matter, I never used such thing in such way before signing-in here. I could use an advice since I'm stuck here, I really don't know what could I do further, I don't know what settings I could change at tripod and where... :oops:
Could you think to another solution?
marcellus

EDIT: I tried to see the thread thru www.anonymization.net and even so I can see the pictures. What this could mean?