[HDR-photo] XMP sidecar file for HDR images - Does it make sense?

Mon Dec 4 12:42:00 EST 2006

I think sidecar files are useful because:

- the algorithmes could improve (e.g. the RAW part)
- Most important are for me the Tonemapping settings

Uwe

--- Bernhard Vogl <bvogl at gmx.at> wrote:

> Ok, now as i found all hdr-photo related mails in my
> spambox, i can also start to particiapte in this
> discussion  ;-)
> 
> * As Mark said, i think it doesn't happen often that
> you re-create a HDR file from brackets. A reason
> could be a change in technology (e.g. ghost/flare
> removal) but normally, you wouldn't care about the
> original images
> * But it be use to store derived EXIF information:
> - anchor exposure
> - DR estimation by EV steps (e.g. a bracket of 3
> images 2EV steps apart -> 4 EV stepped HDR)
> - Camera and Lens
> - color temperature settings (aka white balance)
> * The idea to store some hints about the original
> files is good, but i would strip that down to the
> absolute minimum in a way like PTGui does.
> * Tone mapping parameters make sense, as long as
> multiple operators can store their information and
> retrieve them by some sort of ID.
> 
> For me, the most interesting part would be the
> possibility to have some clues about the "real-world
> information" in the HDR. This is something i am
> badly missing at the moment. It would not only help
> for standardized tonemapping but also for 3D
> integration (IBL).
> 
> Best regards
> Bernhard
> -------- Original-Nachricht --------
> Datum: Sat, 2 Dec 2006 19:29:51 -0500
> Von: listmail at mab3d.com
> An: High Dynamic Range Photography
> <hdr-photo at hdr-photography.com>
> Betreff: Re: [HDR-photo] XMP sidecar file for HDR
> images - Does it make sense?
> 
> > Hi Geraldine,
> > 
> > I think it makes sense only if the XMP includes
> both the composition  
> > information (exposures of source images,
> unchanging EXIF details,  
> > etc.) as well as any tonemapping settings that you
> wish to save with  
> > the file. Like the ACR XMPs, it would have both
> "origin" and  
> > "transformation" information, so that this
> information could be  
> > quickly viewed in a browser and any tonemapping
> re-applied or  
> > adjusted if desired. The catch is that the
> tonemapping information  
> > would be specific to the application and its
> operators, and would  
> > probably be ignored by apps that do not have those
> settings. I don't  
> > see this as a big deal, since there is plenty of
> information in an  
> > ACR XMP that is useless when passed to another RAW
> processor.
> > 
> > The ability to "recreate" an HDR file from its
> source images is  
> > something that would be occasionally useful, but
> once you have the  
> > HDR file and create an XMP file for it, how often
> would "recreation"  
> > be needed? I suppose if you save in a lossy format
> or scale the  
> > image, then going back to the source files would
> help, but otherwise  
> > it would just be a point of information for the
> origins of the file  
> > (kind of like the "auxiliary" information in an
> ACR XMP) and useful  
> > in tagging and cataloging.
> > 
> > Or do you propose an HDR creation method that has
> the flexibility to  
> > add/ remove exposures then recreate the HDR using
> the same response  
> > curve? Personally, I'd like an HDR software that
> can look at the EXIF  
> > data of a pile of photos and automatically match
> them up as "exposure  
> > sets" based on common values (date/ time range,
> plus focal length,  
> > aperture, ISO for finer distinctions).
> > 
> > So, IMHO, it is the tonemapping information that
> would be more useful  
> > on a day-to-day basis - even if it is fairly
> proprietary. That way,  
> > "re-processing" the HDR could be done by starting
> from the previous  
> > results at any time in the future without
> explicitly saving the  
> > settings each time. (Again, just like ACR XMPs)
> > 
> > Putting it all into a single XMP would, I think,
> be the most  
> > efficient way to do all of this, even if a lot of
> the information  
> > would be ignored by other apps or only used in
> rare circumstances.  
> > Its just a modified XML file, right? Fill it with
> all kinds of text  
> > and make reading/ writing to it very flexible so
> that other apps can  
> > adopt it and pass common information between them.
> > 
> > -Mark
> > 
> > On Dec 2, 2006, at 12:40 PM, Geraldine Joffre
> wrote:
> > 
> > > A sidecar XMP file for an HDR image could store
> the following type  
> > > of information:
> > > - Filename and path of the source images used to
> produce the HDR image
> > > - Exposures of the source images
> > > - Relevant exif data hat did not change between
> the exposures  
> > > (focal length,
> > > camera model, owner, ...)
> > > - Color primaries and white point (or name of
> the color profile of  
> > > the source
> > > images)
> > >
> > > The advantage of such XMP file is that the
> metadata of the HDR  
> > > image would be
> > > stored in a standardized way, easily accessible
> by other  
> > > applications -even
> > > directly readable by users-, whereas the various
> HDR formats that  
> > > exist have
> > > all their own way to store metadata, and do not
> always include
> > > photography-related tags anyway.
> > >
> > > The drawback I am seeing though is the potential
> confusion with XMP  
> > > files that
> > > store the settings used to tone map HDR images.
> If an HDR image has  
> > > an XMP
> > > file containing metadata type of information as
> proposed above,  
> > > should the
> > > settings used to tone map this HDR image be
> stored in a separate  
> > > XMP file? If
> > > so, how to differentiate both files to avoid
> confusion? And  
> > > wouldn't the
> > > terminology "sidecar file" loose its sense if
> they are two of them  
> > > associated
> > > with the same HDR image?
> > >
> > > If not, i.e. if both the HDR image metadata and
> the tone mapping  
> > > setting are
> > > included in a single XMP file, could it be
> confusing to have one  
> > > part of the
> > > information that relates to the original HDR
> image (and does not  
> > > change) and
> > > the other part to the processing? And how to
> differentiate between  
> > > an XMP file
> > > that contains both type of information and one
> that contains only  
> > > tone mapping
> > > settings?
> > >
> > > Any comment/input/suggestion will be
> appreciated.
> > >
> > > Thanks,
> > >
> > > Geraldine
> _______________________________________________
> HDR-photo mailing list
> HDR-photo at hdr-photography.com
>
http://www.hdr-photography.com/mailman/listinfo/hdr-photo
> 

Uwe Steinmueller
uwe at outbackphoto.com (Publisher/editor)  www.outbackphoto.com
San Jose, CA/USA  (408 884 3097 Skype: 408 627 4019)         www.photodotcom.com
"Digital Photography Workflow Handbook"
http://www.outbackphoto.com/booklets/booklets.html

NOTE: please also use ustein_outback at yahoo.com as reply email address if you have an urgent matter.