Am 03.08.2012 16:08, schrieb Gerardo - 360Cities:
> We do try to extract the timezone information when available, but the
> timezone is not always present in the exif data.
>
> For example, this is the exif information available on one of Kay's images:
>
> Date/Time Original : 2012:07:09 18:37:08
> Create Date : 2012:07:09 18:37:08
>
> There is no timezone information in those tags. In these cases, when no
> timezone information is available in the exif data, we assume the
> timezone to be UTC.
How would the exif information look if time zone information was in
them? And what tool would I use to add the time zone information to the
exif tags? Can you give a sample exiftool call to get this done? If I
include a suffix like '+02:00' in a call to exiftool to set the
date/time original, it has no effect, only when I modify the create
date. I'd say all the file dates are the OS's business and may mean all
kinds of things, but I'd only resort to them if no exif data can be found.
I feel it's wrong to assume UTC when there is no time zone information.
I think the most obvious guess would be that the time in the exif data
is local time, and if I as the author give a time without timezone
information, I'd expect this time to be shown to the public. Whatever
the content of the tag, it should be interpreted as the information I
intend to have published and not as somthing I am unaware of which needs
processing your end. If I add a time zone, the viewer can display the
timezone information, if I don't, it should just display what it finds
in the tag. Moreover, it is quite irrelevant what time it was at the
viewer's location when the image was taken.
If I upload an image with a time tag of '18:00' and no further
information (like a time zone, however this might be accomplished), I
would expect the site to present my image as taken at the time I have
put into the tag and not at a time generated by
a) the assumption 18:00 is 18:00 UTC
b) the addition of two hours because my (the viewer's) timezone is CEST
which results in a display of 20:00.
If you want to do any trickery along those lines, have a look at the GPS
time stamp, if present (like in my images). The GPS time stamp is, as
far as I know, always UTC.
> To answer the question of what tags do we use to determine the time and
> date on which the panorama was taken, we inspect Create Date first, and
> if it is not available or empty, we inspect Date/Time Original.
I'd say this is debatable: in a panorama date/time original seems the
more obvious choice. The create date may be carried over from the
initial images, or it may not - it might be the moment the panorama
software writes it's product (which is, in fact, the create time of the
panorama), or it may even be some later stage's timestamp.
Thanks for looking into the matter. I hope you'll find a good solution.
I know metadata is a terribly tricky affair, but I think my suggestion
is feasible:
- use date/time original or create time if available
- if no timezone is given, don't try and guess it
- don't adapt the time to the user's timezone
- display the time as given, with or without timezone
This procedure should be straightforward, as it makes no assumptions,
and should produce a result which is preceived as being obvious. Keep it
simple. Trust the user.