Re: [Gimp-user] time saver?

On 11/15/2012 10:13 PM, Jay Smith wrote:
On 11/15/2012 09:48 PM, rmooney wrote:
I enjoy using GIMP, but I'm looking for some help to cut down on my time
watermarking my photos. Currently I am opening a photo, adding text,
text, duplicating the text layer, then moving each layer to the top
left and
bottom right corners. Surely there must be a way to make this process
Any help will be appreciated.

A few years ago I was using Imagemagick for similar watermarking in an
automated batch process.  However, at that time there were some bugs (at
least on our system) that prevented me from getting exactly what I
wanted in terms of text appearance.  However, it was _supposed_ to work
and seemed to for other people.  I suspect it does now (if not then
also) and would work now on our more modern o/s.

When I was using it, it was part of a larger process I ran: making
multiple scaled JPEGs from single TIFFs, and applying the watermark, all
in the same process (without messing up the TIFFs).

We still use Imagemagick for the sizing/scaling to process thousands of
images.  We use a Perl script wrapper to manage the process of running
through hundreds of source directories/folders containing tens of
thousands of images, processing only those that are newer in the source
than the target, and pushing the results to a matching output
directory/folder structure.  (Thus any changed image will get
reprocessed.)  We use a complicated algorithm in the Perl script to
decided the min / max size for each output, with max limitations on
width and/or height (but maintaining original scale), based on the four
target uses and on the physical size of the input object TIFFs (the
TIFFs are all scanned at 100% of size (said from a printing perspective)
at 300 dpi.  The original source objects range from 1x1 inch to 20x20
inches (flatbed scans).  The output sizes need to be "sane" to fit in a
web page that is "sane" and yet takes advantage of whatever size is
available without making the page too wide.  The logic and math of this
hurts my head; among other things, the differences input and output dpi
really warps the math, while you are scaling differently depending upon
various rules.

Anyway, it works great and runs like cat with a bell tied to its tail.
(I'm a cat person, but that would be fun.)  The more RAM and the more
processors the better and if you can control (increase) the number of
simultaneous processes on your machine, you can make it scream.

If we were to go back to watermarking, of course the watermark needs to
be applied _after_ the sizing/scaling, but before saving the JPEG.

You could also set up a "watched folder" type of system into which you
could drop the source images and have a 'cron' job that checks and
processes the folder however often you want.


One added note: By maintaining separate source (input) and target (output) directory structures you ....

- Will never damage your source files.

- If you want to change the watermark on the existing target files for any reason, you simply change the script doing the watermarking, use 'touch' (or similar) to change the dates of either the source or target files so that the process [a Perl wrapper script] will think the source files are newer, and run it again. All your target files will be recreated with the new & improved watermark.

Suggestion: Never delete your target files as a way to tell the process to recreate them. (Instead, move them aside.) I say this because every year I find one or two source files (out of tens of thousands) that have mysteriously been corrupted (though not accessed according to the file internals), the favorite corruption is that the image turns negative. Stray neutrinos?? Anyway, as long as you have your target files, you can at least have something usable even if your source has become bad. [And another good reason for progressive, non-destructive backups stored off premises.]


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]