wich one to use with XCD? MKV or OGM

[mustaneekeri]

I didn't find a simple answer for this so i made a new thread.

which container is safer to use?
other benefits?

[Shootist]

I think this is again a question of compability. I use ogm because most of my encodes are in this container. I never had a cd with a video I could not read. I don`t think there are much differences between ogm and mkv. but only ogm is well supported. that is the only advantage of it.

If you are going to use ac3 with ogm because of the extra space, be aware, there are much problems. the rest works fine.

[zulu]

For the time being i'd say ogm, cause it has built in error protection.
Error protection is also covered in matroska's specs, but its not implemented yet.

[corigan]

I have zero problems with ogm and ac3.

Corigan

[Koepi]

OGM has no error correction or protection built-in. It just can easily cope with errors due to the timestamps/markers the frames/ogg pages have and it's design for streaming (thus it's necessary to allow errors) - it simply skips over them.

Matroska should work ok as well because it has time stamps as well.

Only avi has problems there.

regards
Koepi

[zulu]

@koepi:
thanks for the info, didn't know that.
I just read about ogm and error recovery somewhere, obviously this info wasn't reliable..

[Atamido]

@Koepi: Wouldn't OGM need a form of Error Detection to be able to know if a page was corrupted?

[-=Stan=-]

I tried this test .
Muxed and AVI as Matroska and OGG .
Opened Ultraedit (hex-editor) and deleted a few bytes from each file .
Tried opening the Matroska and OGG in VdubMod .

Ogg loads well but as expected does not show the full sequence .
Matroska fails to load , returns an error .

Tried with another file .
This time same result with OGG .
But Matroska loads this time , however only a part of the sequence .
When trying to play in a media player .

Ogg plays well with some frames removed .
But Matroska fails to show the full sequence , only shows a few frames .

-----------------------

So my question is this - Does Matroska support error-protection ?
If not right now , will it do so in future .

[Koepi]

I'm sure this only happens because of the young age of the filtersets for matroska. They'll get further developed to jump over errors as well. (At least i believe that the design of matroska allows that)

Regards
Koepi

[spyder]

Yes it does allow this. And most likely what he garbled in the file was part of the headers causing the error. Currently or DSF doesn't handle stream errors well. But if the file didn't open at all, this suggests a error in the file headers.

[-=Stan=-]

@spyder

I removed the bytes from middle of a 500 KB file so dont think that i wud have touched the headers .

edit - Typo

[spyder]

Hmmm....strange...did this file open or did the filter refuse playback? And what filter are you using? This should not happen as the filter does not scan that far into the file before beginning playback...

PS: there is planned support for Error Correction not only protection. :)

[-=Stan=-]

This file (500 KB) did open in VdubMod . But only a few frames , i think till the uncorrupted part .
And the filter also did playback only till the part displayed in VdubMod .

The other file (25 MB) which i reduced to 16 MB , refused to open in VdubMod .
I did not try playing the whole file in the player .
But i did start .

Another thing was that the file was unseakable in the player .
However OGG had the same problem .

I was using gabest's latest filters .

[bond]

Originally posted by spyder
there is planned support for Error Correction not only protection. :) is already someone working on it or just planning?

[spyder]

CRC32 error detection was implemented by jcsston. As for ECC, no one is working on it yet.

[Phanton_13]

A good error correction at stream level isn't viable and ned to ad much aditioanl space to the file, when more reliable error correction more space is neded to ad, but if stream alow to detect this error with a proper precision ( ogg has a 32 bit crc value at each page inf: http://www.xiph.org/ogg/vorbis/doc/framing.html ) its possible to implement o simulate a simple error correction for XCD, we only ned to try read de damaged page for ogg as example various times up to optain the correct page, this work well in some cases but we ned a maximun number of tries for desestimate de page or we can have a loop problem, this can be implement only at file spliter in the most cases for video stream formats, we only ned implement the recalling when bad page is detected.

PD. For my OGM and MKV in reality they are a complementary format, but i donk know how wy some people try to make they a adversaries, MKV have the flexibility for will be the best editing format but de OGG is best at streaming and is only enginereed for this. The best similarity for this that I found is AVI vs ASF, the editing of AVI is a far better than the editing of ASF but you only can make streaming with WMV.

[kilg0r3]

I only can confirm the stan's findings. There definitely is problem with skipping corrupted data. BTW it does not seem to matter which splitte filter I use Gabest's or the 'original'

[Atamido]

Originally posted by Phanton_13
A good error correction at stream level isn't viable and would add much aditioanl space to the file. With more reliable error correction more space is needed to add. We hope to be able to make it configurable. IE, if you have a 750MB movie, then create 50MB of ECC. If you have a 780MB move, then 20MB of ECC, and so on. Yes, it takes more space than simple CRC, but it should be fine depending on the amount of space you start with.