Re: [Cellar] Mapping of MOV clean aperture values to Matroska

[Dave Rice <[e-mail]>]

Hi Kieran,

> On Nov 23, 2018, at 4:56 AM, Kieran O Leary <[e-mail]> wrote:
> 
> Hi,
> 
> Would I be correct in saying that ideally, Clean Aperture values in a
> MOV/MPEG-4 file should map to the PixelCrop element in Matroska?

Yes, they conceptually do the same thing but I’m a little confused about how to use the vert/horiz offset fractions of the clap atom to convert to the pixel crop atom. I tried in both QuickTime 7 and X, and while QuickTime X uses the cleanAperature w/h fractions to present a cropped image, QuickTime 7 doesn’t but appears to adjust the aspect ratio according to those values. QuickTime X does seem impacted by non-zero offset values not in a way that I consider coherent. Do you have any examples of un-centered aperatures? I was hoping that IMX files would, does the ones I have don’t use any aperature.

> I'm asking about this in regards to this ffmpeg ticket -
> https://trac.ffmpeg.org/ticket/7437
> As it appears that FFmpeg does not currently read the values in the
> clap atom and it also does not map them to Matorksa files when
> remuxing. It seems to make the most sense to map them to PixelCrop,
> but I'd like to hear it from folks more knowledgable than me.

Seems reasonable to me, but I’m not exactly certain what the mapping would be.

> Another associated issue with the display of these kinds of values is
> with VLC as it appears that
> 
> 1) PixelCrop values in Matroska are not supported by VLC at the moment
> - https://trac.videolan.org/vlc/ticket/21192
> 2) Clean aperture values are not supported by VLC at the moment
> https://trac.videolan.org/vlc/ticket/21179

Sounds like these tickets are in my priority order as well.
Dave Rice