[FFmpeg-devel] [PATCH 2/2] Fix side channels when XCh extension is present

[Kostya]

On Sun, Jul 11, 2010 at 09:19:48AM +0100, Nick Brereton wrote:
> On Sun, 2010-07-11 at 09:45 +0300, Kostya wrote:
> > First, this is not unified diff, so it's not a patch in strict sense :)
> 
> Sorry about that - fixed! (& attached)

Well, if you start forgetting to attach patches that means you're ready
to be proper FFmpeg developer ;)
 
> > And can you guarantee it's always correct mapping for channels (sorry,
> > don't have specs at hand) i.e. it does not depend on amode?
> 
> s->channel_order_tab is a pointer to a row in the static channel mapping
> tables at the top of the file, which already takes amode into account.
> 
> As this problem only occurs for DTS-ES matrix signals when there is a
> discrete back-channel decoded (XCh) I think that the only significant
> cases are amode=9 (i.e. 5 channels) with or without LFE. I have tested
> samples of both of these cases which I think work correctly.
> Unfortunately I'm not aware of a (publicly available) spec that covers
> this extension.

I have some access to the specs, I'll look it up tomorrow.
 
> It would also work for 4 channels + xch (although I'm not sure that's
> used in practice); and also some of the 6 main channel cases - however I
> doubt that's useful.

I just see that some amodes lack SL and SR channels, dunno yet about XCh
amode so I just wanted to check it.

Probably patch ok, I'll wait for Benjamin's words.

> Nick