XML/XMP question for Hert

[PhilBurton]

Hert,

I've been looking at XMP files created by PSu and by Lightroom, and files created by PSu and then modified by Lightroom. I was wondering if there is any significance to the following:

A PSu XMP file has about 12 rdf:description entries under the main rdf:RDF section. In Lightroom, there is only one rdf:description.

If Lightroom reads in a PSu XMP file and then writes that file out (e.g. after modifications), all the various items in each rdf entry in the PSu file are combined into the one rdf entry in the Lightroom XMP file.

If PSu reads in that Lightroom XMP file and then writes it out, all the entries will again be arranged in the approximately 12 rdf entries in the PSu style.

Phil

[fbungarz]

Sorry to chime in...
I can only speculate, but I believe the fact that Adobe and PSu write out the same block of XMP may be essentially a "design flaw" how XMP works.
XMP is basically a language that tells programs how to interpret image metadata. The fact that it is eXtensible means that anyone who understands its syntax can use it to define different metadata "fields". This, unfortunately does not necessarily mean that any program will understand and correctly interpret these "fields" correctly.
Typically this is not a problem unless you use a program to modify a certain block of data. Then a program that essentially does not "understand" a certain metadata section will ideally ignore it, but in reality may simply rewrite it and "correct" the syntax according to what it "should" look like...
I presume that is what is happening here. Adobe does not completely "understand" the metadata that PSu writes and thus simply overwrites it with what it presumes is the correct format. As long as PSu can still correctly read this, this is minor. Some programs, apparently PhaseOne is a good example, however, will simply delete blocks of metadata (or perhaps re-write it so that it becomes unintelligible to PSu). If that happens the result is loosing that information...

Reading a bit about it all on Phil Harvey's Exiftool site, it seems that most metadata formats are essentially flawed:
http://www.sno.phy.queensu.ca/~phil/exi ... dards.html

So in theory, XMP was designed for metadata interchangeability, in practice it doesn't actually seem to work all that well.

Frank

[Hert]

Lr writes its namespaces in one go, psu writes namespaces there where applicable. Xml wise they are both fine. I think psu's XMP is better readable.

[PhilBurton]

Sorry to chime in...
I can only speculate, but I believe the fact that Adobe and PSu write out the same block of XMP may be essentially a "design flaw" how XMP works.
XMP is basically a language that tells programs how to interpret image metadata. The fact that it is eXtensible means that anyone who understands its syntax can use it to define different metadata "fields". This, unfortunately does not necessarily mean that any program will understand and correctly interpret these "fields" correctly.
Typically this is not a problem unless you use a program to modify a certain block of data. Then a program that essentially does not "understand" a certain metadata section will ideally ignore it, but in reality may simply rewrite it and "correct" the syntax according to what it "should" look like...
I presume that is what is happening here. Adobe does not completely "understand" the metadata that PSu writes and thus simply overwrites it with what it presumes is the correct format. As long as PSu can still correctly read this, this is minor. Some programs, apparently PhaseOne is a good example, however, will simply delete blocks of metadata (or perhaps re-write it so that it becomes unintelligible to PSu). If that happens the result is loosing that information...

Reading a bit about it all on Phil Harvey's Exiftool site, it seems that most metadata formats are essentially flawed:
http://www.sno.phy.queensu.ca/~phil/exi ... dards.html

So in theory, XMP was designed for metadata interchangeability, in practice it doesn't actually seem to work all that well.

Frank

Frank,

Perhaps Adobe should have held "Implementors' Conferences" every few years to address these issues.

Phil

[PhilBurton]

Lr writes its namespaces in one go, psu writes namespaces there where applicable. Xml wise they are both fine. I think psu's XMP is better readable.

Hert,

Totally agree here. After seeing the PSu XMP file, I was shocked when I saw that LR creates only one rdf section.