i've been recently working on another gallery generator for my personnal
website. I've almost completed a working clone of RDFPic, the picture
commenter from the W3, using Ruby plus a homemade EXIF parsing library
and a GTK2 interface. and i've started working on the generator itself.
The idead being this generator is a kind of tool �-la make that starting
from source folder, a set of templates and a destination folder,
generates a bunch of html files and jpeg thumbnail (plus optionally,
rescaled big size file to avoid putting 3Mpix files on the net). This
"set" of files is supposed to be ready to be put on the net.
I know you fellows are quite brilliant when it comes to thinking, so
I'll expose my problem. Apologies if I sound too "demanding" and boring.
I want to find a way to generate thumbnail _and_ associate them to their
original big jpeg file so as to be able to be sure the thumbnail is up
to date ie the original big jpeg file hasn't changed. What I want to
allow is: a modification of any of the metadata (except say the
orientation) should not force me to rebuild a thumbnail.
I was thinking of using a jpeg COMment inside the thumbnail to indicate
the checksum of the original jpeg file (excluding the metadata).
Using this system i also wanted to "track" the thumbnail in case the
users modifies it source tree (so that my gallery will find the
thumbnail wherever it is in the destination tree and link/move it to
good new destination place).
What would you think of such a system ?
This also means that to find a given thumbnail for a big jpeg file, I
must know the destination path in some way. I'm also thinking of adding
the thumb path to the main JPEG file.
Super. I installed Sun's Java to try out that W3C java hack but was
greatly disappointed of the sucky Java-GUI and Java's constant poor
performance. So I could wipe Java off my disk, but I still needa similar
application that handle the Photo-RDF.
Re: Photo thumbnail naming and tracing
Charles McCathieNevile <[email protected]>,
Fri, 20 Jun 2003 13:33:21 -0400 (EDT)
This seems cool. There is some stuff that Nick Kew and Jim Ley did on making
checksums of parts of documents - so you could have a checksum over
everything except the metadata, if you don't already have that.
Dealing with things that move around is just a pain. Which is why cool URI's
don't change. (On the other hand not having a high-bandwidth server issued
free at birth is a pain too...)
I'd be interested in pointing people to this stuff from a W3C project if