-vcodec

[betonpfeiler]

Hey Everyone!
Here comes my first question for this mailinglist:

If I want to encode a file, for testing purposes, several times with the
same codec. Is there kind of a "force reencode" command or does "-vcodec
<codec>" really reencode the file?
But, in my tests I wasn't sure to spot any difference between several
generations. If that's true... FFMpeg is an just unbelievable...
Thanks for answering my simple question, but I wasn't able to find anything
on this topic in the Internet..

Bye,
J
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user

[Andy Civil]

On 2010-11-17 10:45 AM, kalle ralle wrote:

> Hey Everyone!
> Here comes my first question for this mailinglist:
>
> If I want to encode a file, for testing purposes, several times with the
> same codec. Is there kind of a "force reencode" command or does "-vcodec
> <codec>" really reencode the file?
> But, in my tests I wasn't sure to spot any difference between several
> generations. If that's true... FFMpeg is an just unbelievable...
> Thanks for answering my simple question, but I wasn't able to find anything
> on this topic in the Internet..

This is just a novice observation, but I feel that although we think of
compression as 'lossy', it may not be lossy if the file we are compressing has
been decompressed from the identical compression scheme.

What I mean is, that a lossy compression will find the best* approximation to
the source material... but when the source has already been decompressed by the
same scheme, the compressor can find a perfect match; we know that, because it
was compressed that way before.

As a real-world example, suppose you had to create someone's image using a
mosaic of coloured stones, you could make an approximation. But if you were
asked to make an image of a mosaic of coloured stones... you could do it perfectly.

(*with so much effort going into defining what 'best' means, LOL!)

IMHO!

--
Andy
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user

[James Darnley]

On 17/11/2010, kalle ralle <[hidden email]> wrote:

> Hey Everyone!
> Here comes my first question for this mailinglist:
>
> If I want to encode a file, for testing purposes, several times with the
> same codec. Is there kind of a "force reencode" command or does "-vcodec
> <codec>" really reencode the file?
> But, in my tests I wasn't sure to spot any difference between several
> generations. If that's true... FFMpeg is an just unbelievable...
> Thanks for answering my simple question, but I wasn't able to find anything
> on this topic in the Internet..

Yes, it really does do a full encode!  Even if the input and output
codecs are the same!  If you want ffmpeg to copy then you would use
-vcodec copy.

If you don't see generational loss with ffmpeg's lossy codecs then you
aren't doing enough generations or the loss in each one is too small.
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user

[Jason Garrett-Glaser]

On Wed, Nov 17, 2010 at 8:18 AM, Andy Civil <[hidden email]> wrote:

> On 2010-11-17 10:45 AM, kalle ralle wrote:
>>
>> Hey Everyone!
>> Here comes my first question for this mailinglist:
>>
>> If I want to encode a file, for testing purposes, several times with the
>> same codec. Is there kind of a "force reencode" command or does "-vcodec
>> <codec>" really reencode the file?
>> But, in my tests I wasn't sure to spot any difference between several
>> generations. If that's true... FFMpeg is an just unbelievable...
>> Thanks for answering my simple question, but I wasn't able to find
>> anything
>> on this topic in the Internet..
>
> This is just a novice observation, but I feel that although we think of
> compression as 'lossy', it may not be lossy if the file we are compressing
> has been decompressed from the identical compression scheme.
>
> What I mean is, that a lossy compression will find the best* approximation
> to the source material... but when the source has already been decompressed
> by the same scheme, the compressor can find a perfect match; we know that,
> because it was compressed that way before.

Nope, doesn't work that way.  Yes in theory it could, in practice it doesn't.

Dark Shikari
_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user

[betonpfeiler]

> ----------------------------------------------------------------------
>
> Message: 1
> Date: Wed, 17 Nov 2010 17:47:06 +0100
> From: James Darnley<[hidden email]>
> Subject: Re: [FFmpeg-user] -vcodec
> To: FFmpeg user questions and RTFMs<[hidden email]>
> Message-ID:
> <AANLkTimJhExuEoomBejOV4DK5=[hidden email]>
> Content-Type: text/plain; charset=UTF-8
>
> On 17/11/2010, kalle ralle<[hidden email]>  wrote:
>    
>> Hey Everyone!
>> Here comes my first question for this mailinglist:
>>
>> If I want to encode a file, for testing purposes, several times with the
>> same codec. Is there kind of a "force reencode" command or does "-vcodec
>> <codec>" really reencode the file?
>> But, in my tests I wasn't sure to spot any difference between several
>> generations. If that's true... FFMpeg is an just unbelievable...
>> Thanks for answering my simple question, but I wasn't able to find anything
>> on this topic in the Internet..
>>      
> Yes, it really does do a full encode!  Even if the input and output
> codecs are the same!  If you want ffmpeg to copy then you would use
> -vcodec copy.
>
> If you don't see generational loss with ffmpeg's lossy codecs then you
> aren't doing enough generations or the loss in each one is too small.
>
>    

Well, I did a 7 Generationstest with. Because I couldn't believe that
the FFMpeg really did a reencode everytime I tested it with an Youtube
Clip, 5 Geneartions at 0.5M mpeg-2. The first really showed artifacts,
every other Generation stayed the same. I'll run some tests with
objective metrics soon... Still can't believe it. Any other encoder I've
used showed generational loss in a big way...
But if FFmpeg relly works so perfect... Hats off to everyone involved in
this project!!

_______________________________________________
ffmpeg-user mailing list
[hidden email]
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user