New Tool Kutter - Page 2

rds_correia · #21 09-03-2005, 07:42 PM

Quote:

Originally Posted by kwag

I'll look into adding 64 bit file size support, and then you can feed it your whole hard drive

Cool

Quote:

Originally Posted by kwag

Until then, please don't use files larger that 2GB

No problem

rds_correia · #22 09-03-2005, 07:47 PM

Quote:

Originally Posted by kwag

Did you try feeding the original VOB to DVDLab, and see if you get the same message

Yep, it seems they are identical to the source because all VOB sources on my disk produce such a message when I try to load them

.
Let me tell you that I have never ever tried to add a VOBfile to a DVDLab project.
Usually I only do M2Vs and AC3s and have DVDLab mux them and I never get such a message.

kwag · #23 09-03-2005, 07:54 PM

Quote:

Originally Posted by rds_correia

Could this be a DVDLab vs Noob problem at 1h38am

Or is DVDLab having difficulty in accepting the chunks?

May very well be a DVDLab bug

Have you tried using MPEG Validator to analyze your source, and then the chunks

http://sr2.mytempdir.com/144314

-kwag

kwag · #24 09-03-2005, 08:03 PM

Or make a quick test Rui

After you chop all your pieces (I post three VOBs on the example), and assuming your original VOB is named VTS_01_2.VOB, do this on a command prompt:

copy /B VTS_01_2.VOB.kutter.1.VOB + VTS_01_2.VOB.kutter.1.VOB + VTS_01_2.VOB.kutter.1.VOB joined.vob

Now the file "joined.vob" should be an identical bit by bit to the file VTS_01_2.VOB

-kwag

rds_correia · #25 09-03-2005, 08:03 PM

Stand by.
Working on many tests to see if it has to do with the chunks or with DVDLab.
I'll be back in '5.
Cheers

rds_correia · #26 09-03-2005, 08:27 PM

I did many tests but none suited my needs.
So I went for one last and very long test.
I added 2 vobfiles with 2 full movies that I just extracted to my HDD with DVDDecrypter.
And this time there were no messages and the muxing went well from the begining until the end.
I just burnt it to a DVD-RW and it's playing fine in the living room.
So the issues that I had earlier must come from the chunks done by Kutter.
Or maybe I'm missing something...

kwag · #27 09-03-2005, 09:49 PM

Hmm, I just did the copy test, and the final joined VOB is identical to the source

Code:

C:\kutter>fc joined.vob h:\RED_PLANET\VIDEO_TS\VTS_01_2.VOB
Comparing files joined.vob and H:\RED_PLANET\VIDEO_TS\VTS_01_2.VOB
FC: no differences encountered


C:\kutter>

That means there's a perfect match bit for bit between the files.
Unless the headers are buggy, and I'm not correctly tagging the cuts on the individual pieces.
I'll do more tests.

-kwag

Zyphon · #28 09-04-2005, 02:16 AM

@rds_correia & Kwag

Thanks guys for the extensive testing you have both done, I am sorry I haven't had time to test myself but I was backing up one of my DVD's

It is a shame about the 2Gb limitation of PB, I wish the developer of this programming language would fix this issuse as it holds back the excellent developers here who use it such as Karl and Andrej.

Hopefully that 64bit binary Inc supplied will come in use.

I hope to run some tests myself today.

About the Open GOPs in DVD-Lab, I get this notification also when adding a .mpv stream made by CCE from my DVD sources but it doesn't affect the authoring and I have no probs when I watch the backups on my SAP.

I just thought that info might come in useful.

Prodater64 · #29 09-04-2005, 05:10 AM

Quote:

Originally Posted by kwag

Or make a quick test Rui

After you chop all your pieces (I post three VOBs on the example), and assuming your original VOB is named VTS_01_2.VOB, do this on a command prompt:

copy /B VTS_01_2.VOB.kutter.1.VOB + VTS_01_2.VOB.kutter.1.VOB + VTS_01_2.VOB.kutter.1.VOB joined.vob

Now the file "joined.vob" should be an identical bit by bit to the file VTS_01_2.VOB

-kwag

When you apply this copy /b command, files are joined but headers are not rewrited, then you will have a file that plays fine but time info will be wrong. Later when you reencode a such video, a-v will be async.
It was my experience a time ago while programed a tool.
I dont know if in this case it could be important.

Edited: I worked with mpgs and not with vobs.

Boulder · #30 09-04-2005, 07:47 AM

Quote:

Originally Posted by kwag

Hmm, I just did the copy test, and the final joined VOB is identical to the source

Code:

C:\kutter>fc joined.vob h:\RED_PLANET\VIDEO_TS\VTS_01_2.VOB
Comparing files joined.vob and H:\RED_PLANET\VIDEO_TS\VTS_01_2.VOB
FC: no differences encountered


C:\kutter>

That means there's a perfect match bit for bit between the files.
Unless the headers are buggy, and I'm not correctly tagging the cuts on the individual pieces.
I'll do more tests.

-kwag

Did you try fc /B ?

rds_correia · #31 09-04-2005, 10:07 AM

Quote:

Originally Posted by kwag

That means there's a perfect match bit for bit between the files.
Unless the headers are buggy, and I'm not correctly tagging the cuts on the individual pieces.
I'll do more tests.

Just let me know if I can help

Cheers

kwag · #32 09-04-2005, 11:52 AM

@Boulder,

I also tried the FC /B. Result
C:\kutter>fc /B joined.vob h:\RED_PLANET\VIDEO_TS\VTS_01_2.VOB
Comparing files joined.vob and H:\RED_PLANET\VIDEO_TS\VTS_01_2.VOB
FC: no differences encountered

I suspect I might have a bug in the end header on the chunks.
As a matter of fact, this has to be the problem, because the way it's suppposed to work is that each chunk is a valid MPEG file with start and end headers. So a join of all chunks shouldn't be the same as the original, because each chunk has to have added ending headers.
I'll fix this today

@All,
Thanks for all the feedback

-kwag

incredible · #33 09-04-2005, 12:12 PM

You won't loose informations by cutting if encoding the mpegs using SEQ Header at every Gop. Thats normally the "default" as mpeg was also meant to be a streaming format.
And here Kutter should ONLY cut at SeqHeaders.

That header can be easely parsed as it begins using .....

Code:

If ReadFile(#File_Temp, p_filename) ; open file for read-only
    tempL = ReadLong() ; read in the first 32 bit
    If p_forceHeaderSearch = #False And tempL <> $B3010000 And tempL <> $BA010000
      ProcedureReturn #False ; return if not a common MPEG-file and no forceHeaderSearch is set
    ElseIf tempL <> $B3010000
      Repeat ; search bytealigned for a (possible) sequence-header (simply straight forward, no chunk-jumps here)
        tempL = ReadLong()
        FileSeek(Loc()-3)
      Until Eof(#File_Temp) Or tempL = $B3010000 ; until fileend reached or sequenceheader found
      If tempL = $B3010000
        FileSeek(Loc() + 3) ; set the filepointer back
      Else
        ProcedureReturn #False ; there's no sequence header in this file
      EndIf
    EndIf
Endif

I used that in Packshot and Mencalc to get the needed Stream Infos when importing mpeg2/1 Files.
The "Parser" searches in the opened file till he founds a sequence header, then do a FileSeek(Loc()-3) so you get again in front of the $B3010000 Long value and THERE you do your cuts!

If only ONE Seq Header at the beginning of the source is found then you MUST parse it completely and add it to your cutted target chunks at their very beginnings!

I Missed something:
You cant cut mpegs just by bytes! You have to cut at an I Frame startpoint.
Do look at the (if available) mpeg2schnitt sources, there you can see how its being done

kwag · #34 09-04-2005, 02:09 PM

Quote:

Originally Posted by incredible

You won't loose informations by cutting if encoding the mpegs using SEQ Header at every Gop. Thats normally the "default" as mpeg was also meant to be a streaming format.

That's exactly what I'm doing, Andrej

Quote:

And here Kutter should ONLY cut at SeqHeaders.

It does

Quote:

That header can be easely parsed as it begins using .....

Code:

If ReadFile(#File_Temp, p_filename) ; open file for read-only
    tempL = ReadLong() ; read in the first 32 bit
    If p_forceHeaderSearch = #False And tempL <> $B3010000 And tempL <> $BA010000
      ProcedureReturn #False ; return if not a common MPEG-file and no forceHeaderSearch is set
    ElseIf tempL <> $B3010000
      Repeat ; search bytealigned for a (possible) sequence-header (simply straight forward, no chunk-jumps here)
        tempL = ReadLong()
        FileSeek(Loc()-3)
      Until Eof(#File_Temp) Or tempL = $B3010000 ; until fileend reached or sequenceheader found
      If tempL = $B3010000
        FileSeek(Loc() + 3) ; set the filepointer back
      Else
        ProcedureReturn #False ; there's no sequence header in this file
      EndIf
    EndIf
Endif

I used that in Packshot and Mencalc to get the needed Stream Infos when importing mpeg2/1 Files.
The "Parser" searches in the opened file till he founds a sequence header, then do a FileSeek(Loc()-3) so you get again in front of the $B3010000 Long value and THERE you do your cuts!

If only ONE Seq Header at the beginning of the source is found then you MUST parse it completely and add it to your cutted target chunks at their very beginnings!

I Missed something:
You cant cut mpegs just by bytes! You have to cut at an I Frame startpoint.
Do look at the (if available) mpeg2schnitt sources, there you can see how its being done

Very good code, although I already had my implementation, which is completely different

Actually this one you posted is a little simpler than the one I had, but if it ain't broke, don't fix it

Now there's something that just came to my mind. What happens if an actual header sequence pattern shows up in the data (video) part

What I mean is, in standard data communications, if a control/command pattern is part of the data, it's usualy preceded by some escape sequence.
This way, you can differentiate from actual headers or from data.
So what happens in an MPEG stream, when a header byte pattern is part of the data

If it's not preceeded by some escape sequence, this can (and will!) throw off any decoder, because it thinks it's a start of a sequence header, when in fact it's actual data

Anyone ever thought about this, or did I miss something in the MPEG protocol

-kwag

rds_correia · #35 09-04-2005, 02:48 PM

Man!!
I know you're probably right but I'm affraid I didn't catch up a word of what you wrote in the last paragraph

.
Cheers

kwag · #36 09-04-2005, 03:03 PM

Don't worry about it Rui

-kwag

incredible · #37 09-04-2005, 03:18 PM

I do think these 4 bytes in their count are something like protected!?

Cause if not, such a parser would sometimes put out garbage - and that never happened to me.

kwag · #38 09-04-2005, 04:45 PM

Quote:

Originally Posted by incredible

I do think these 4 bytes in their count are something like protected!?

But it CAN be a valid four bytes in a video stream, so there must be some identification bytes (escape sequence) to differentiate commands vs. data.

Quote:

Cause if not, such a parser would sometimes put out garbage - and that never happened to me.

Maybe all decoders are depending on the following couple of bytes after the start header, which are 0x2d, 0x01.
That would make the pattern 00, 00 ,01, B3, 2D, 01 much more difficult to produce in actual data, but it still CAN happen

-kwag

kwag · #39 09-04-2005, 05:40 PM

@Rui,

Did you try importing .M2V file chunks into DVDLab

I just tried importing several .M2V chunks into TMPGEnc DVD Author, and there were no errors.
The original VOB file was demuxed to .M2V with TMPGEnc tools to produce the file, which I then chopped up with Kutter.

-kwag

rds_correia · #40 09-04-2005, 06:11 PM

No Karl, unfortunately I didn't try that.
I only tried it on VOBfiles.
Today I'm encoding.
Tomorrow I may find time to test it on M2Vs

.
Cheers