[FFmpeg-user] more sound madness

[Bouke / VideoToolShed]

> On 07 Mar 2020, at 18:05, Ted Park <kumowoon1025 at gmail.com> wrote:
> 
> Hi,
> 
>> Hi Ted,
>> Sorry, I have no clue what you mean here, could you please elaborate?
> I am not sure if this is accessible with ffmpeg, but in the track header, usually only tracks with different alternate_group, or 0, can play at the same time.
> 
>> Nope, moving the  does not make any difference.
>> Did you try this yourself? Does it work for you?
> I mean, yes, but it works either way. But the source I used was a single mxf with multiple tracks, I didn’t try multiple inputs or a bwf.

Hi Ted,
Did you try it in Premiere, and got all the channels? Or open the output in old QT pro, and look at the properties.
The last channels are disabled. If I enable them and save the file (destructive), a new MOOV atom is added and the current one is flagged as FREE.
Then Premiere will accept it as it ‘should’ be.
(Warning, Premiere has  a nasty habit to ‘remember’ files. Once it has seen a specific file, it will use the properties of the first import, even when you change it and re-import.)

> 
>>>> Stream mapping:
>>>> Stream #0:0 -> #0:0 (copy)
>>>> Stream #0:1 -> #0:1 (pcm_s16be (native) -> pcm_s24le (native))
>>>> Stream #1:0 -> #0:2 (pcm_s24le (native) -> pcm_s24le (native))
>>>> Stream #1:0 -> #0:3 (pcm_s24le (native) -> pcm_s24le (native))
>>>> Press [q] to stop, [?] for help
>>>> -map_channel is forwarded to lavfi similarly to -af pan=0x4|c0=c1.
>>>> [pan @ 0x7fb020e00a00] Pure channel mapping detected: 1
>>>> -map_channel is forwarded to lavfi similarly to -af pan=0x4|c0=c8.
>>>> [pan @ 0x7fb020d04740] Pure channel mapping detected: 8
>>>> -map_channel is forwarded to lavfi similarly to -af pan=0x4|c0=c9.
>>>> [pan @ 0x7fb020f1d200] Pure channel mapping detected: 9
>>> It shouldn’t matter if audio channels come before others.
>> trust me, it does matter. First input file gets the audio enabled, all other gets muted in metadata.
> Oh right… But I mean they don’t matter if your tracks aren’t randomly disabled. I still would have thought at least one would get chosen.
> When you say muted in metadata, do you literally mean they have their relative volume set to 0 per track?
> Either way, I’m pretty sure it is because some channels overlap between the two inputs, I think this filter is what is actually being attempted, give it a try.
> % ffmpeg -i /Volumes/Data/test/C0092.MP4 -i /Volumes/Data/test/068.WAV -filter_complex “[0:1]pan=1c|c0=c1[o1];[1:0]pan=1c|c0=c9[o2];[1:0]pan=1c|c0=c10[o3]” -map “[o1]" -map “[o2]" -map “[o3]"
> -c:a pcm_s24le -map 0:0 -c:v copy -shortest /Volumes/Data/test/last2muted.mov -y

This gives me exactly the same issue.

Now, I can live with all the sound channels in one track, only reason I’m pursuing this is that I can’t remap more than about 8 channels into a single poly track.
(See my post from a few days ago…)

Thanks,
Bouke

>> there is no ‘normal’ in my world, and the Mov format accepts virtually unlimited channels in virtual unlimited tracks in virtual unlimited combinations, so that ‘should’ not be an issue.
> The manual panning sets all tracks as mono, at channel 0. The mov format has very little restrictions, but what is supported by software is a different story.
> 
>>> or how the extra channels are mapped wrt gain. Exploding the multitrack recording into separate wave files first might make things simpler.
>> 
>> SImpler in what way?
>> Sorry, but merging two files ‘should’ be possible IMHO, and I’m fighting on doing this for a couple of days now, so ‘simple’ is not in my book anymore.
> Yes I agree, it was just an alternative because what should be working doesn’t seem to be. It’s simpler since you don’t have to worry about the channels being in a known layout for one thing.
> 
> Regards,
> Ted Park
> 
> _______________________________________________
> ffmpeg-user mailing list
> ffmpeg-user at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-user
> 
> To unsubscribe, visit link above, or email
> ffmpeg-user-request at ffmpeg.org with subject "unsubscribe".