Subsurface media import problems

[Berthold Stoeger]

On Sonntag, 6. September 2020 20:19:13 CEST Berthold Stoeger via subsurface 
wrote:

> > So i can't achieve to modifiy the date and time value read by
> > 
> >       sub-surface. Using "Attribute changer" (https://www.petges.lu/)
> >       does not work. I can't figure where is hidden the value read by
> >       subsurface.
> 
> The code responsible for reading out the metadata is found in core/
> metadata.cpp. There are multiple ways that the timestamp is extracted, so it
> depends on your video file.

Eric sent me a sample video two weeks ago and I finally found time to look at 
it.

Here's the output of the "atom" (I'm still dumbfounded by the fact that 
"atoms" contain "subatoms") tree as seen by our parser (note that we descend 
only into selected "atoms"):

atom: ftyp
atom: mdat
atom: moov
-atom: mvhd
-atom: udta
--atom: SDLN
--atom: smrd
--atom: smta
-atom: meta
--atom: hdlr
--atom: keys
                key mdta com.android.version
--atom: ilst
-atom: trak
--atom: tkhd
--atom: mdia
---atom: mdhd
                timestamp: 1599227282
---atom: hdlr
---atom: minf
-atom: trak
--atom: tkhd
--atom: mdia
---atom: mdhd
                timestamp: 1599227282
---atom: hdlr
---atom: minf

The timestamp, which corresponds to Fri, 04 Sep 2020 13:48:02 UTC, is where 
you would expect it, viz. in the mdhd "atom". Does anybody know where else a 
timestamp could be hidden? Should we add an option to prefer file-dates over 
meta-data? I'm not a videographer, so no idea about video manipulation 
software, sorry. EXIFtools can add XMP data, however that should be in an UUID 
atom, which is not found here - at least not at the top level.

Berthold