| Sign in: |
| Members log in here with your user name and password to access the your admin page and other special features. |
|
|
|
|
|| SportsShooter.com: Member Message Board
New EXIF Metadata Standard Announced
Dave Amorde, Photographer
 |
Lake Forest | CA | USA | Posted: 6:12 PM on 09.26.08 |
->> http://www.ddj.com/architect/210603847
"Adobe, Apple, Canon, Microsoft, Nokia, and Sony look to set digital photo standard"
Excerpt: " 'Lack of metadata interoperability has led to significant frustration for both consumer and pro photographers, and our companies have spent considerable resources trying to deal with the problem,' said Josh Weisberg, chairman and founder of the Metadata Working Group and director of Microsoft's Rich Media Group."
Looks like we may finally see the day when Photoshop automatically recognizes new RAW formats, and browswers recognize future compressed media formats. |
|
Walter Calahan, Photographer
|
Westminster | MD | USA | Posted: 6:40 PM on 09.26.08 |
->> 'bout time.
thanks |
|
Landon Finch, Photographer
|
Colorado Springs | CO | USA | Posted: 9:41 PM on 09.26.08 |
| ->> Where's Nikon? |
|
Primoz Jeroncic, Photographer
|
Kranj | SI | Slovenia | Posted: 5:52 AM on 09.27.08 |
->> Somehow I think this has nothing to do with PS recognizing new RAW formats, but just about plain EXIF. IPTC is more or less standardized already. Even though it's very very lose standard.
Well... to be honest, I though till now, EXIF is standardized too, but obviously it isn't. |
|
Christopher Blunck, Photographer
|
Hyattsville | MD | USA | Posted: 6:05 PM on 09.29.08 |
->> The closest thing I've found to a standard format is XMP encoded in the APP13 segment of a JPG file. XMP was started by Adobe and is a logical extension to the work they did in the 90s on IPTC/EXIF. It's an XML representation of IPTC metadata fields that's pretty easy to work with once you read it out of the JPG data.
The only beef I've seen with XMP is that the schema that describes the XML is not easy to find on the Internet. It's easy enough to read the XML from the JPG and parse it after you see several examples, but actually coding to the standard (as represented in XSD) is difficult. |
|
Israel Shirk, Photographer, Assistant
|
McCall | ID | US | Posted: 7:03 PM on 09.29.08 |
->> Chris-
Agreed.
It always frustrates me when I work with a RAW image in one application and then use another to archive/sort, and it deletes any data related to the image... The combination of Photoshop and Photo Mechanic conflicts way too much. One will write something somewhere in the headers or XMP sidecar, then the other will either move or re-arrange it, and suddenly it doesn't work.
Grrrrr.
Even the APP13 segment does get tweaky sometimes if you play with it too much - some programs won't recognize something you customized with a script, and then think, "It's corrupt! Overwrite EVERYTHING in that segment." Oy. Thanks.
But it's usually pretty easy to recover from, as compared with working with RAW recipes where you have to re-do everything manually...
Hopefully this'll be the fix! |
|
Christopher Blunck, Photographer
|
Hyattsville | MD | USA | Posted: 9:18 PM on 09.29.08 |
->> @Israel Shirk:
The JPG standard is pretty difficult. I tried writing some code to pull out the byte[] that holds the EXIF/IPTC and had a really hard time figuring out how to do it without scanning the entire file. The section markers in JPG are duplicated for the different image streams inside the file and there doesn't seem to be a way to say "this is my main image" vs "this is a thumbnail" and "this is another thumbnail".
With each image stream metadata was present. I examined JPG files produced by PhotoShop, Elements, Aperture 1.5, Aperture 2.0, and LR2 and found inconsistencies between all of them. I can't remember the particulars because it was a month or so ago but the immediate items that jumped out were: the main image was sometimes first and other times it was last, and the IPTC/EXIF was near the thumbnail and not the main image.
I also remember all of the complexity of the actual image. Is it JFIF? Or is it JPEG? I'm forgetting a lot of the details because I focused on the metadata aspect but oh man was it complex!
Walking away from that project gave me a much bigger appreciation for exactly what lives inside a JPG file. It's a lot more than just binary image data! |
|
Allen Murabayashi, Photographer
|
New York | NY | USA | Posted: 9:51 PM on 09.29.08 |
->> i just skimmed the specification.
the main feature is that they have created precedence in dealing with a file that might have multiple meta data standards embedded within in.
for example, if you used an old program to create meta data with IPTC-IIM, then you pulled it into a newer program that wrote meta data using XMP, this standard says "default to IPTC-IIM." this is good and bad as you can extrapolate.
it also gives character limits for certain fields (e.g. keywords can't occupy more than 64 bytes, ratings should be floating points from -1.0 to 5.0).
it is designed to be very consumer oriented, so there's not a ton of benefit for pros right now other than informing software developers of how to deal with multiple meta data blocks in the same image. |
|
Israel Shirk, Photographer, Assistant
|
McCall | ID | US | Posted: 12:54 AM on 09.30.08 |
->> Clarification:
There is a specific array of settings which works as a workaround if you're using Photo Mechanic and photoshop, requiring you to tweak both photoshop and photo mechanic. However, by default, Photo Mechanic will delete any information stored in an XMP sidecar by Photoshop if you modify the file in photo mechanic afterwards. Kirk from Camera Bits e-mailed me that there is a way to change this behavior, but I haven't had any luck with it. I'll try again tomorrow and post it if I have any luck.
At the same time, one really great thing about Photo Mechanic is that it takes care of most of the captioning and keywording standards for you right now... I never really appreciated it until I started having to figure out what standards might apply to IPTC... All the stuff with IPTC-IIM, different standards per country, etc... Photo mechanic turns that from a nightmare into a walk in the park!
Chris-
Amen. I just worked with the IPTC and EXIF info too; once I started looking at the possibilities for how to have my scripts batch-resize images... Eventually I figured out how to sort it out but can't remember any of it now. If you (or anyone else) interested I still have those scripts sitting around - they're in Python, which is pretty readable if you don't already know it.
I looked at the JPEG/JFIF/JPEG2000 standards and thought to myself, "I would like to KEEP my sanity!" And quickly figured out a couple of libraries to take care of that for me! Thank God.
And for the sake of making this searchable: "Working with JPEG, RAW, TIFF, IPTC, EXIF, metadata, extraction, insertion, standards, xmp, xml, help, file header." |
|
Christopher Blunck, Photographer
|
Hyattsville | MD | USA | Posted: 10:14 AM on 09.30.08 |
->> @Israel Shirk:
I'm definitely interested in your scripts!
Here's the problem I had: my JPG images are not actually rotated but the orientation flag is set. Preview (in OSX) interprets this flag properly but FireFox, Safari, and IE do not. As a result I have to rotate my images prior to publishing in order for them to be viewed properly.
Rotating the images using Jimi, java.awt, Sanselan, JAI all stripped the APP13 metadata from the image once they performed the rotation. The same applies when they scaled the image. I whole-heartedly need that APP13 metadata embedded in my JPG files because it's used extensively by down-stream services (e.g. PhotoShelter, Flickr).
My approach to solving the problem was to look for the APP13 segment markers in the original image, pull them out, and then insert them into the rotated or resized image. I prototyped it in Python but ran into all the problems we discussed above (multiple image streams, inconsistent locations for the image streams, my lack of understanding of the JPG spec).
I ultimately turned to the Sanselan project and worked with one of the OSS developers to get IPTC "writing" into the project roadmap. Those folks know a lot more about the JPG standard than I do and they already have a great library for reading the IPTC metadata out. Hopefully they'll have write support built soon and I'll be able to just incorporate that logic into my publishing tool.
Anyhow ... it'd be great if you'd send along your python script. My email address is: cblunck@dcsportsbox.com. |
|
Israel Shirk, Photographer, Assistant
|
McCall | ID | US | Posted: 2:22 PM on 09.30.08 |
->> Here's a workaround for Photo Mechanic that will keep you from deleting your photoshop recipes:
http://www.sportsshooter.com/brianshirk/pm_xmp/pages/1.html
Again, not a clean solution in the least... It just keeps Photo Mechanic from using the sidecar at all. So then you have two sets of XMP (the embedded and sidecar), which in certain situations (near and dear to my heart) makes life very miserable.
Go team! Standardize! Play nicely at last!
Chris-
I'll look around and see if I can find them... I'm actually about to do some updating to my website which will require me to update them too. |
|
Israel Shirk, Photographer, Assistant
|
McCall | ID | US | Posted: 3:21 PM on 09.30.08 |
->> And another update:
It looks like the default settings for Photo Mechanic changed at some point to prevent the problem. My old settings from a while ago just never got updated.
Nothing to see here, carry on... |
|
Return to --> Message Board Main Index
|