On Friday 29 June 2007 15:25, Patrick Shanahan wrote:
I process >25k photos/year and prefer a usb card-reader.
I also prefer using a card reader.
My photos are sorted: Year.Month.Day.Event/Scene processed and archived (x2) to dvd+
I use a similar scheme that worked out very nicely for me - purely using the
file system. That's really cool because I can use plain and simple tools for
viewing or to look up photos. Or even to back everything up to USB disk,
laptop, over the network or whatever. Database based applications are more
complex to handle in that regard.
It basically looks like this:
/work/photo/
orig/
dvd01/
2005-08-30-Prag/
2005-09-10-Barcelona/
2005-10-16-Pegnitztal/
dvd02/
...
sort/
Germany/
Nuernberg/
Castles/
Rabenstein/
Greifenstein/
Journeys/
Prag/
Barcelona/
import/
2007-06-28-Hack-Week/
The basic structure should be pretty obvious: I have a
directory /work/photo/orig/ where I store the originals as they come from the
camera. I create a new subdirectory for each photo outing, prefixed with date
and with a descriptive name (2005-09-10-Barcelona). I group them into "dvdXY"
subdirectories so I can easily burn them on DVD; I check with KDirStat every
once in a while and when the latest dvdXY directory grows beyond 4 GB, I
create a new one and move the last couple of subdirectories to that new one
until the old one gets below 4 GB.
/work/photo/sort is where I put the post-processed photos. There are
subdirectories for logical grouping so I can see all photos that belong to
any topic together -- say, all photos I shot in the Nuernberg old town.
There is also /work/photos/sort/import for photos waiting to be post-processed
and sorted.
And here comes a twist: I have also a directory /work/photos/by-date/. It
mirrors the structure of /work/phots/orig (without the dvdXY in between) so I
can quickly see the photos by photo outing - but the post-processed ones, not
the unprocessed originals from the camera:
/work/photos/by-date/
2005-08-30-Prag/
2005-09-10-Barcelona/
2005-10-16-Pegnitztal/
Each of those subdirs holds symlinks (!) for each photo from that outing to
the photo with the same name somewhere under /work/photos/sort/:
/work/photos/by-date/2005-09-10-Barcelona/20050910-1648-img_8431.jpg ->
/work/photos/sort/Journeys/Barcelona/20050910-1648-img_8431.jpg
As you can see, the photos are renamed: I prefix them with the shooting date
and time from the EXIF data. I do that immediately upon image import to
orig/, so even the photos there bear that kind of name. But it's simple to
restore the original file name: Simply remove the date and time prefix.
That timestamp within the camera name has several advantages:
- All cameras from all vendors wrap their image counter around after 9999:
After shooting img_9999.jpg, the next will be img_0001.jpg. That means that
file names are no longer unique after 9999 photos. (And yes, it sucks, but
it's specified like that). With the date+time prefix, this does not matter
any more.
- You can use several cameras from the same maker that each use the same
naming scheme and dump them into one directory without name clashes. That's
particularly nice when throwing together several people's photos from a party
or a wedding. And they will even be sorted by time among all cameras when
viewing with Konqueor, Kuickshow, Gwenview or other file system based tools.
- Photos made with different cameras over time will be sorted chronologically
in the sort/ subdirectories. It's always simple to see the latest photos,
disregarding vendor-specific naming schemes (Canon uses img_1234.jpg, Sony
uses dsc01234.jpg (and that leading 0 is a fake, it won't be used in the
image counter; it will wrap around from dsc09999.jpg to dsc00001.jpg).
Of course, I use scripts to do most of that stuff: I have a
script "photo_rename" (based on "jhead" with a little Perl around it) that
renames the photos based on their shooting date and time. And backward, too,
if requested. And not double-adding any date+time to files that already have
them.
I also have a script "photo_clone_by_date" that looks at the orig/ hierarchy,
creates a corresponding subdirectory below by-date/, searches below sort/ for
files with the same name and creates symlinks (runs ~12 seconds for ~8000
photos in ~200 subdirs below orig/).
And I have a script "photo_rsync" (actually little more than a
glorified "rsync" call) that helps copying all that stuff from my computer to
my USB disk or to my laptop or to my Archos image viewer (copying files
rather than creating symlinks for FAT file systems like on that Archos image
viewer / MP3 player).
Now I know that my approach might not be for everyone. But it might give some
ideas.
If anybody wants to experiment, here are the scripts (hereby GPLed):
http://www.suse.de/~sh/photo-util/
CU
--
Stefan Hundhammer